- Email: [email protected]
Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management
G Model
ARTICLE IN PRESS
RESUS-7001; No. of Pages 7
Resuscitation xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management夽,夽夽 Byung Kook Lee a , Kyung Woon Jeung a,∗ , Yong Hun Jung a , Dong Hun Lee a , Sung Min Lee a , Yong Soo Cho a , Tag Heo a , Jong Geun Yun b , Yong Il Min a a b
Department of Emergency Medicine, Chonnam National University Hospital, 42 Jebong-ro, Donggu, Gwangju, Republic of Korea Department of Emergency Medical Services, Honam University, 417 Eodeung-daero, Gwangsangu, Gwangju, Republic of Korea
a r t i c l e
i n f o
Article history: Received 28 August 2016 Received in revised form 28 November 2016 Accepted 4 December 2016 Keywords: Heart arrest Induced hypothermia Body temperature regulation Prognosis
a b s t r a c t Aim of the study: Studies examining associations between time to target temperature and outcomes in cardiac arrest patients who underwent targeted temperature management (TTM) have shown inconsistent results. We examined these associations separately for time from restoration of spontaneous circulation to TTM initiation (pre-induction time) and time from TTM initiation to target temperature (induction time). Furthermore, we examined whether critical time thresholds exist if there is an association. Methods: This was a single-centre retrospective observational study including adult cardiac arrest patients treated with TTM from 2008 to 2015. We tested the associations of pre-induction time and induction time with outcomes at hospital discharge using multivariate logistic regression analysis. We then performed additional multivariate analyses, each with the significant timing variable at different binary cutoffs. Results: A total of 515 patients were analysed. At hospital discharge, 357 patients (69.3%) were alive, of whom 161 (31.3%) had a favourable neurologic outcome. In multivariate analysis, a shorter preinduction time was independently associated with a favourable neurologic outcome (odds ratio [OR], 1.110; 95% confidence interval [CI], 1.025–1.202), whereas the induction time was not (OR, 0.954; 95% CI, 0.852–1.067). We found two pre-induction time thresholds (120 and 360 min) that were associated with neurologic outcome. Conclusion: We found that a shorter pre-induction time was independently associated with a favorable neurologic outcome at hospital discharge, whereas induction time was not. We also found two time thresholds at 120 and 360 min, after which initiation of cooling was associated with a worse neurologic outcome. © 2016 Published by Elsevier Ireland Ltd.
Introduction
Abbreviations: TTM, targeted temperature management; ROSC, restoration of spontaneous circulation; CNUH, Chonnam national university hospital; ECMO, extracorporeal membrane oxygenation; CPR, cardiopulmonary resuscitation; GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; CPC, Cerebral Performance Categories; OR, odds ratio; CI, confidence interval. 夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at http://dx.doi.org/10.1016/j.resuscitation.2016.12.002. 夽夽 Chonnam National University Hospital Institutional Review Board Protocol No. CNUH-2016-104. ∗ Corresponding author. Fax: +82 62 228 7417. E-mail addresses: [email protected] (B.K. Lee), [email protected] (K.W. Jeung), [email protected] (Y.H. Jung), [email protected] (D.H. Lee), [email protected] (S.M. Lee), [email protected] (Y.S. Cho), [email protected] (T. Heo), [email protected] (J.G. Yun), [email protected] (Y.I. Min).
Targeted temperature management (TTM) has been widely recommended for the treatment of comatose cardiac arrest survivors since the publication of two landmark papers on therapeutic hypothermia.1,2 Despite a number of studies regarding the TTM after cardiac arrest, many questions remain unanswered. Among them, the relationship between timing of cooling and outcome is an unresolved issue that bears important clinical implications. Animal studies have consistently shown that earlier cooling results in greater benefits from this therapy.3–5 However, human studies have shown inconsistent results on the impact of earlier cooling.6–14 It appears reasonable to examine the association between time to target temperature, which is the time interval from restoration
http://dx.doi.org/10.1016/j.resuscitation.2016.12.002 0300-9572/© 2016 Published by Elsevier Ireland Ltd.
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model RESUS-7001; No. of Pages 7
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
2
36–36.5 ◦ C. All patients undergoing TTM received continuous intravenous midazolam and remifentanil (or fentanyl) until at least 72 h after cardiac arrest. A neuromuscular blocking agent was administered to control shivering on an as-needed basis. All other aspects of patient management were at the discretion of the treating physicians.
of spontaneous circulation (ROSC) to achievement of target temperature, and outcomes in order to understand how the timing of cooling impacts outcomes of comatose cardiac arrest survivors. The time to target temperature can be divided into two time intervals: pre-induction time, which is the time interval from ROSC to TTM initiation, and induction time, the time interval from TTM initiation to achievement of target temperature. According to the animal studies showing benefits of earlier cooling, a shorter pre-induction time is expected to be associated with favourable outcomes.3–5 In contrast, clinical studies suggest that a shorter induction time is associated with worse outcomes.11,14 Thus, regarding the preinduction time and induction time, the associations with outcomes can be in the opposite direction, and the direction and significance of the association between time to target temperature and outcomes may vary depending on the relative contribution of the two time intervals to outcomes, thus resulting in inconsistent results in human studies. In light of this, it may be more reasonable to distinguish between pre-induction time and induction time in exploring the association between the timing of cooling and outcomes. To our knowledge, few studies examined this association separately for the preinduction and induction times.11,14 In the present study, we sought to examine the associations of pre-induction time and induction time with survival to hospital discharge and neurologic outcome at hospital discharge. Furthermore, we sought to examine whether critical time thresholds exist if there is an association. We hypothesised that a shorter pre-induction time would be associated with favourable outcomes while a shorter induction time would be associated with unfavourable outcomes.
The following data were obtained from hospital records: age, sex, comorbidities, first monitored rhythm, aetiology of cardiac arrest, location of cardiac arrest, presence of a witness on collapse, bystander cardiopulmonary resuscitation (CPR), time to ROSC, transfer status (initially presented at CNUH versus initially presented at another hospital and transferred to CNUH), Glasgow Coma Scale (GCS) score after ROSC, initial temperature (recorded on hospital admission), TTM device, pre-induction time, induction time, duration of rewarming, sequential organ failure assessment (SOFA) score within the first 24 h after admission,15 vital status at hospital discharge (alive or dead), and neurologic outcome at hospital discharge. For the calculation of pre-induction and induction times, the initiation of TTM was defined as the time when the first attempt of cooling was made. Neurologic outcome was assessed using the Glasgow–Pittsburgh Cerebral Performance Categories (CPC) scale at discharge and recorded as CPC 1 (good performance), CPC 2 (moderate disability), CPC 3 (severe disability), CPC 4 (vegetative state), or CPC 5 (brain death or death).16 The primary outcome was an unfavourable neurological outcome, defined as CPC 3–5 at hospital discharge. The secondary outcome was in-hospital mortality.
Methods
Statistical analysis
Study design and population
Categorical variables were presented as frequencies and percentages. Comparisons of categorical variables were performed using 2 or Fisher exact tests, as appropriate. Continuous variables were presented as median values with interquartile ranges because all continuous variables showed non-normal distribution. The Mann–Whitney U test was conducted for comparisons of continuous variables. The Kruskal–Wallis test was performed to compare induction times among the cooling devices. General linear model analysis was performed to identify the interaction between transfer status and location of cardiac arrest. Multivariate logistic regression analysis was used to examine the association between the timing of cooling and outcomes after adjusting for potential confounders. All variables with p < 0.2 in univariate analyses were included in the multivariate regression model. The multicollinearity between variables was assessed before modelling. Backward selection was used to obtain the final model. First, we entered the timing variables into the model as continuous variables. Second, to examine whether a significant time threshold value could be determined, we performed additional multivariate logistic regression analyses, each with the significant timing variable at different binary cutoffs (60-min increments from 60 to 420 min of preinduction time). Data were analysed using PASW/SPSSTM software, version 18 (IBM Inc., Chicago, IL, USA). A two-sided significance level of 0.05 was used for statistical significance.
This was a retrospective observational cohort study of adult comatose cardiac arrest survivors treated with TTM at Chonnam National University Hospital (CNUH) in Gwangju, Korea, from January 2008 to December 2015. The Institutional Review Board of CNUH approved this study. Cardiac arrest patients over 18 years of age who underwent TTM were included. Patients were excluded if (1) TTM was interrupted owing to transfer to another facility or haemodynamic instability, (2) they died before the target temperature was reached, (3) they were treated with a different TTM protocol, or (4) extracorporeal membrane oxygenation was applied during post-cardiac arrest care. TTM protocol No cooling attempt was made in the pre-hospital phase, and thus TTM was started after arrival at our hospital in all patients of this study. According to our protocol, TTM was considered for all nontraumatic cardiac arrest survivors who could not obey commands. Patients were not eligible for TTM if they had intracranial haemorrhage, active bleeding, known terminal illness, or a poor pre-arrest neurologic status. Cooling was initiated as soon as possible with ice packs, intravenous cold saline, and one of the ® following three TTM devices: Arctic Sun Energy Transfer PadsTM ® (Medivance Corp, Louisville, KY, USA); Blanketrol II (Cincinnati ® Subzero Products, Cincinnati, OH, USA); or COOLGARD3000 Thermal Regulation System (Alsius Corporation, Irvine, CA, USA). The target temperature of 33 ◦ C was achieved and then maintained for 24 h using one of the TTM devices. Upon completion of the 24-h maintenance phase, active rewarming was attempted at a target rate of 0.25 ◦ C–0.5 ◦ C h−1 until the body temperature reached
Data collection
Results A total of 636 adult cardiac arrest patients were treated with TTM during the study period. Of these, 121 patients were excluded as shown in Fig. 1. Thus, 515 patients were included in this study. At hospital discharge, 357 patients (69.3%) were alive, of whom 161 (31.3%) had a favourable neurologic outcome. Clinical characteristics stratified by outcomes are shown in Table 1. Patients
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
Unfavourable neurologic outcome (n = 354)
p-Value
Survivor (n = 357)
Nonsurvivor (n = 158)
p-Value
60.0 (47.0–70.0) 341 (66.2)
52.0 (42.0–61.0) 117 (72.7)
64.0 (51.0–72.0) 224 (63.3)
<0.001 0.037
58.0 (46.0–69.0) 235 (65.8)
65.0 (51.0–73.0) 106 (67.1)
<0.001 0.780
75 (14.6) 39 (7.6) 201 (39.0) 135 (26.2) 30 (5.8) 54 (10.5) 38 (7.4) 12 (2.3) 402 (78.1)
27 (16.8) 77 (6.8) 48 (29.8) 21 (13.0) 3 (1.9) 9 (5.6) 6 (3.7) 2 (1.2) 124 (77.0)
48 (13.6) 28 (7.9) 153 (43.2) 114 (32.2) 27 (7.6) 45 (12.7) 32 (9.0) 10 (2.8) 278 (78.5)
0.338 0.668 0.004 <0.001 0.010 0.014 0.033 0.270 0.701 0.937
49 (13.7) 26 (7.3) 131 (36.7) 80 (22.4) 19 (5.3) 28 (7.8) 27 (7.6) 9 (2.5) 282 (79.0)
26 (16.5) 13 (8.2) 70 (44.3) 55 (34.8) 11 (7.0) 26 (16.5) 11 (7.0) 3 (1.9) 120 (75.9)
0.418 0.709 0.103 0.003 0.464 0.003 0.810 0.666 0.442 0.044
418 (81.2) 97 (18.8)
131 (81.4) 30 (18.6)
287 (81.1) 67 (18.9)
298 (83.5) 59 (16.5)
120 (75.9) 38 (24.1)
153 (29.7) 111 (21.6) 217 (42.1) 34 (6.6)
97 (60.2) 31 (19.3) 24 (14.9) 9 (5.6)
56 (15.8) 80 (22.6) 193 (54.5) 25 (7.1)
129 (36.1) 75 (21.0) 127 (35.6) 26 (7.3)
24 (15.2) 36 (22.8) 90 (57.0) 8 (5.1)
282 (54.8) 117 (22.7) 74 (14.4) 38 (7.4) 4 (0.8) 388 (75.3) 254 (49.3) 28.0 (17.0–40.0) 3 (3–4) 36.0 (35.0–36.7)
132 (82.0) 13 (8.1) 5 (3.1) 11 (6.8) 0 (0.0) 140 (87.0) 89 (55.3) 21.0 (13.0–30.0) 4 (3–6) 36.4 (35.9–37.0)
150 (42.4) 104 (29.4) 69 (19.5) 27 (7.6) 4 (1.1) 248 (70.1) 165 (46.6) 30.0 (20.0–40.0) 3 (3–3) 36.0 (34.7–36.5)
215 (60.2) 68 (19.0) 50 (14.0) 22 (6.2) 2 (0.6) 277 (77.6) 180 (50.4) 25.0 (15.0–36.5) 3 (3–4) 36.0 (35.3–36.8)
67 (42.4) 49 (31.0) 24 (15.2) 16 (10.1) 2 (1.3) 111 (70.3) 74 (46.8) 32.0 (20.0–42.5) 3 (3–3) 35.8 (34.6–36.5)
137 (26.6) 135 (26.2) 243 (47.2) 215 (150–300) 2.5 (1.5–4.0) 12.0 (10.4–13.6), 482c 10 (7–12)
39 (24.2) 48 (29.8) 74 (46.0) 202 (150–300) 3.0 (2.0–5.0) 12.0 (10.0–13.0) 8 (6–10)
98 (27.7) 87 (24.6) 169 (47.7) 225 (155–302) 2.0 (1.2–3.5) 12.0 (11.0–14.0), 321c 11 (8–12)
90 (25.2) 92 (25.8) 175 (49.0) 230 (159–300) 2.8 (1.8–4.0) 12.0 (10.5 − 13.0) 9 (6–11)
47 (29.7) 43 (27.2) 68 (430) 198 (145–300) 2.0 (1.0–3.5) 12.0 (10.0–14.0), 125c 11 (10–14)
<0.001
<0.001
<0.001
<0.001 0.068 <0.001 <0.001 <0.001 0.421
0.105 <0.001 0.029 <0.001
0.003
0.075 0.453 0.001 <0.001 0.001 0.414
0.143 <0.001 0.549 <0.001
CAD, coronary artery disease; CVA, cerebrovascular accident; VF, ventricular fibrillation; VT, ventricular tachycardia; PEA, pulseless electrical activity; CPR, cardiopulmonary resuscitation; ROSC, restoration of spontaneous circulation; SOFA, sequential organ failure assessment. a Patients initially presented to another hospital and then transferred to Chonnam National University Hospital. b Of the 418 out-of-hospital cardiac arrest patients, 90 initially presented at CNUH, whereas 328 initially presented at another hospital and were transferred to CNUH. Among the 97 in-hospital cardiac arrest patients, 23 experienced cardiac arrest in CNUH, whereas 74 experienced cardiac arrest in another hospital and were transferred to CNUH. In general linear model univariate analysis, no interaction was found between transfer status and location of cardiac arrest for either the neurologic outcome (p = 0.407) or in-hospital mortality (p = 0.804). c Number of patients for whom rewarming duration was available.
ARTICLE IN PRESS
Comorbidities CAD Heart failure Hypertension Diabetes Pulmonary disease Renal disease CVA Hepatic disease Transfera Locationb Out-of-hospital In-hospital First monitored rhythm VF/pulseless VT PEA Asystole Unknown Aetiology Cardiac Other medical Asphyxia Drug overdose Drowning Witnessed collapse Bystander CPR Time to ROSC (minutes) Glasgow coma scale Initial temperature (◦ C) Cooling device Blanket Endovascular Hydrogel pad Pre-induction time (minutes) Induction time (hours) Rewarming duration (hours) SOFA score
Favourable neurologic outcome (n = 161)
G Model
RESUS-7001; No. of Pages 7
Age (years) Male sex
Total (n = 515)
B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
Table 1 Demographic and arrest characteristics stratified by outcomes.
3
G Model RESUS-7001; No. of Pages 7 4
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
Fig. 1. Schematic diagram showing the number of patients included in the present study. TTM, targeted temperature management; ECMO, extracorporeal membrane oxygenation.
Fig. 2. Neurologic outcome according to the pre-induction time (A) and the induction time (B).
with favourable neurologic outcome were younger and predominantly male. They had a lower incidence of comorbidities, were more likely to have a shockable rhythm, were more likely to have a cardiac aetiology, had a higher incidence of witnessed collapse, had a shorter time to ROSC, had a higher GCS score after ROSC, had a higher initial temperature, had a shorter duration of rewarming, and had a lower SOFA score compared with patients with
unfavourable neurologic outcome. Fig. 2 shows the neurologic outcome of the patients according to the pre-induction and induction times. The pre-induction time did not differ between favourable and unfavourable neurologic outcome groups, whereas the induction time was longer in patients with favorable neurologic outcome. The pre-induction time was significantly longer in transferred patients (235 [166–312] min) than in nontransferred patients (157
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model
ARTICLE IN PRESS
RESUS-7001; No. of Pages 7
B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
5
Table 2 Multivariate logistic regression analysis for an unfavourable neurologic outcome and in-hospital mortality.
Age (years) Diabetes Pulmonary disease Previous CVA Shockable rhythm Cardiac aetiology Time to ROSC (every 5 min) GCS Initial body temperature (◦ C) SOFA score Pre-induction time (every 30 min) Induction time (hours)
Unfavourable neurologic outcome OR (95% CI)
p-Value
In-hospital mortality OR (95% CI)
p-Value
1.027 (1.008–1.046) 2.051 (1.039–4.167) 7.497 (1.160–48.446) 6.863 (1.835–25.660) 0.222 (0.114–0.435) 0.266 (0.136–0.519) 1.290 (1.163–1.432) 0.546 (0.435–0.686) 0.798 (0.630–1.011) 1.099 (1.004–1.203) 1.110 (1.025–1.202) 0.954 (0.852–1.067)
0.006 0.039 0.034 0.004 <0.001 <0.001 <0.001 <0.001 0.062 0.041 0.010 0.408
1.002 (0.985–1.020) 1.190 (0.660–2.753) NA NA 0.347 (0.188–0.639) 0.800 (0.466–1.374) 1.114 (1.030–1.204) 0.630 (0.475–0.835) 1.092 (0.925–1.289) 1.225 (1.137–1.319) 0.965 (0.905–1.030) 0.949 (0.848–1.062)
0.801 0.413
0.001 0.419 0.007 0.001 0.300 <0.001 0.286 0.360
Pre-induction time and induction time were entered into the model as continuous variables. OR, odds ratio; CI, confidence interval; CVA, cerebrovascular accident; ROSC, restoration of spontaneous circulation; GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; NA, not available.
[115–237] min; p < 0.001). The induction time differed significantly according to the TTM device used (3.3 [2.0–5.4] h for the blanket, 2.0 [1.0–4.0] h for the endovascular cooling catheter, and 2.3 [1.3–3.3] h for the hydrogel pad; p < 0.001). Survivors were younger and had a lower incidence of comorbidities. They were more likely to have a shockable rhythm and cardiac aetiology and were more likely to have had an out-of-hospital cardiac arrest (Table 1). Survivors had a shorter time to ROSC, higher GCS score, higher initial temperature, and lower SOFA score compared with nonsurvivors. Pre-induction time did not differ between survivors and nonsurvivors, whereas induction time was significantly longer in survivors. Association of pre-induction time and induction time with neurologic outcome Table 2 shows the result of multivariate logistic regression analysis with pre-induction time and induction time entered as continuous variables. A shorter pre-induction time was independently associated with a favourable neurologic outcome (odds ratio [OR], 1.110; 95% confidence interval [CI], 1.025–1.202), whereas induction time was no longer significantly associated with the neurologic outcome (OR, 0.954; 95% CI, 0.852–1.067). When the patients were divided into two groups according to the time to ROSC (median 28 min), the association between pre-induction time and neurologic outcome was significant only in patients with shorter time to ROSC (Fig. 3). Additional multivariate analyses performed with the pre-induction time at different binary thresholds revealed threshold effects at the pre-induction times of 120 and 360 min (Table 3). Pre-induction times longer than 120 min (OR, 2.264; 95% CI, 1.026–4.995) and longer than 360 min (OR, 2.694; 95% CI, 1.139–6.372) were independently associated with an unfavourable neurologic outcome compared with pre-induction times of ≤120 min and ≤360 min, respectively (Table 3). Table 3 Multivariate logistic models of association between pre-induction time at different binary cutoffs and unfavourable neurologic outcome. Pre-induction time (min)
OR (95% CI)
p-Value
>60 >120 >180 >240 >300 >360 >420
0.801 (0.039–16.551) 2.264 (1.026–4.995) 1.585 (0.906–2.774) 1.265 (0.716–2.235) 1.319 (0.686–2.537) 2.694 (1.139–6.372) 2.497 (0.961–6.490)
0.886 0.043 0.107 0.419 0.407 0.024 0.060
OR, odds ratio; CI, confidence interval.
Fig. 3. Multivariate logistic regression analysis for poor neurologic outcome in patients with time to restoration of spontaneous circulation ≤28 min (A) and >28 min (B). SOFA, sequential organ failure assessment; GCS, Glasgow coma scale, CI, confidence interval.
Association of pre-induction time and induction time with in-hospital mortality Multivariate analysis showed that shockable rhythm (OR, 0.360; 95% CI, 0.196–0.661), time to ROSC (OR, 1.014; 95% CI, 1.002–1.027), GCS (OR, 0.622; 95% CI, 0.469–0.824), and SOFA score (OR, 1.225; 95% CI, 1.137–1.320) were independently associated with in-hospital mortality. Neither the pre-induction time nor the induction time was associated with in-hospital mortality in the multivariate analysis (Table 2).
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model RESUS-7001; No. of Pages 7
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
6
Discussion In this study, we found that a shorter pre-induction time was independently associated with a favourable neurologic outcome at hospital discharge, whereas induction time had no independent association with neurologic outcome. We also found two preinduction time thresholds (120 and 360 min), after which initiation of cooling was associated with a worse neurologic outcome. Studies have suggested that a more precipitous temperature drop induced by cooling is associated with worse outcomes,11,14 reflecting a loss of the ability to resist cooling owing to impaired thermoregulatory control. In a multicentre observational study including 321 comatose cardiac arrest patients, patients with an induction time >300 min had a higher likelihood of a favourable neurologic outcome compared with those with an induction time <120 min.14 In line with these studies,11,14 greater heat generation, which impedes the cooling process, was associated with reduced ischaemic injury and better neurologic outcome in a study that evaluated patient heat generation during TTM after cardiac arrest.17 In our study, the induction time was longer in patients with a favourable neurologic outcome, although this association did not reach statistical significance in multivariate analysis. Nonetheless, our study, along with the previous studies showing shorter induction time in patients with worse outcomes, suggests that preinduction time and induction time should be analysed separately when exploring the association between timing of cooling and outcomes. Similar to our study, Perman et al. and Haugk et al. distinguished the pre-induction time and induction time in exploring the association between time to target temperature and outcomes.11,14 However, in contrast to our study, pre-induction time did not differ significantly between patients with favourable versus unfavourable neurologic outcomes in their studies. The reasons for the difference between our findings and those of these studies are not readily apparent. One possible explanation is that compared with these studies, the pre-induction time in our study was relatively long and had a wide range, which might be adequate to reveal the threshold effects present at pre-induction time >120 min. Our study included a large proportion of patients transferred from district or secondary hospitals in a large rural area, as well as patients who experienced cardiac arrest at our hospital; this resulted in a long pre-induction time with a wide range. In the present study, the association between pre-induction time and neurologic outcome was significant in patients with time to ROSC ≤28 min, whereas this association was not significant in patients with time to ROSC >28 min. This finding is in line with the results of several studies suggesting loss of efficacy of cooling in patients with prolonged time to ROSC.18,19 In a study that compared neurologic outcome between two target temperature ranges (32.0–33.5 ◦ C versus 34.0–35.0 ◦ C) using a multicentre observational registry of post-cardiac arrest patients treated with TTM, cooling was associated with improved neurologic recovery in patients with time to ROSC ≤30 min, whereas it was not in patients with time to ROSC >30 min.18 To our knowledge, no study has explored whether a time threshold could be determined for the neuroprotection conferred by cooling. In the present study, we found two pre-induction time thresholds at 120 and 360 min. This finding should be interpreted in light of some limitations. We did not find a significant threshold at pre-induction time <120 or >360 min. However, considering the relatively small numbers of patients with pre-induction time <120 or >360 min in our study, further studies are required to determine whether a critical threshold exists at pre-induction time <120 or >360 min. Experimental studies have indicated that hypothermia mitigates neuronal injury even when initiated several hours after
ischaemia.20,21 However, despite a great number of clinical studies on TTM after cardiac arrest, the time limit within which cooling should be initiated to achieve its therapeutic benefit remains unknown. Currently, pre-hospital cooling is not widely adopted owing to a lack of firm evidence supporting its use. Furthermore, TTM is limited mostly to a few large hospitals in many countries.22–24 Thus, many cardiac arrest patients are transferred to a TTM-capable facility with the ensuing loss of valuable time and are often admitted to the TTM-capable facility after several hours have elapsed since the cardiac arrest. Information on the window of opportunity is important, particularly to optimise the care of these patients. Our study did not directly address the window of opportunity for hypothermia. Nonetheless, the presence of a pre-induction time threshold at 360 min suggests that the time window of the initiation of cooling should be at most 6 h after ROSC. In this study, pre-induction time was independently associated with neurologic outcome, but it had no significant association with in-hospital mortality. Although this finding is in line with that of a previous study in which early cooling improved neurologic outcome rather than survival,4 this finding should be interpreted cautiously, because our study did not evaluate long-term survival and patients with poor neurologic outcome at discharge may have a higher risk of long-term mortality. In our study, the presence of a witness on collapse and bystander CPR, both of which have traditionally been regarded as important variables influencing cardiac arrest outcome, were not independently associated with neurologic outcome. The reason for this finding is not readily apparent, but the adjustment of time to ROSC might have weakened the effects of these variables in our multivariate model. Consistent with our finding, Komatsu et al. investigated the association of several prehospital and in-hospital factors, including the presence of witness and bystander CPR with outcome in 227 cardiac arrest survivors, and reported that time to ROSC was the only independent factor related to neurologic outcome.25 On the other hand, comorbidities including pulmonary disease and previous CVA were independently associated with neurologic outcome, with adjusted OR of 7.4 and 7.0, respectively. The effect size of these variables should be assessed through further study, since confidence intervals of these variables were large in our multivariate model. Moreover, a recent study suggests that comorbidity itself is not independently associated with neurologic outcome.26 Our study has several limitations. First, this was a single-centre study. Thus, our results require further confirmation by other studies. Second, generalisability to other country settings is uncertain. In Korea, physicians are generally not permitted to withdraw lifesustaining therapies unless the patient is pronounced brain dead. Thus, treatment was not withdrawn for any patient included in this study. This may explain, at least in part, the high proportion of patients with an unfavourable neurologic outcome in the present study. Third, it was a retrospective observational study, and thus, selection bias could have been present despite our efforts to control for this. Fourth, because of its observational nature, our study identified only association and could not determine a causal relationship. Fifth, several factors that could possibly influence the efficacy of TTM, including number of ice packs used, amount of cold saline infused, body habitus, proportion of body surface area covered by cooling pads, and administration of anti-shivering medications, were not fully investigated in our study. Sixth, despite our efforts, there may have been unmeasured confounding factors. For example, transferred patients might not only experience delay in TTM initiation, but could also have received suboptimal care before admission to our hospital. Furthermore, patients with more severe injury might have been transferred to our hospital for TTM. Seventh, pre-induction time involves systemic issues and a number of
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model RESUS-7001; No. of Pages 7
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
factors can influence the length of pre-induction time. However, our study did not address this issue. Conclusions In this study, we found that a shorter pre-induction time was independently associated with a favourable neurologic outcome at hospital discharge whereas induction time had no independent association with outcome. We also found two pre-induction time thresholds at 120 and 360 min, after which initiation of cooling was associated with a worse neurologic outcome. Our results suggest that pre-induction time and induction time should be analysed separately when exploring the association between the timing of cooling and outcomes. Conflict of interest statement All authors have no potential conflicts of interest to disclose. References 1. Bernard SA, Gray TW, Buist MD, et al. Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. N Engl J Med 2002;346:557–63. 2. Hypothermia after Cardiac Arrest Study Group. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Engl J Med 2002;346:549–56. 3. Che D, Li L, Kopil CM, Liu Z, Guo W, Neumar RW. Impact of therapeutic hypothermia onset and duration on survival, neurologic function, and neurodegeneration after cardiac arrest. Crit Care Med 2011;39:1423–30. 4. Kuboyama K, Safar P, Radovsky A, Tisherman SA, Stezoski SW, Alexander H. Delay in cooling negates the beneficial effect of mild resuscitative cerebral hypothermia after cardiac arrest in dogs: a prospective, randomized study. Crit Care Med 1993;21:1348–58. 5. Nozari A, Safar P, Stezoski SW, et al. Critical time window for intra-arrest cooling with cold saline flush in a dog model of cardiopulmonary resuscitation. Circulation 2006;113:2690–6. 6. Chiota NA, Freeman WD, Barrett K. Earlier hypothermia attainment is associated with improved outcomes after cardiac arrest. J Vasc Interv Neurol 2011;4:14–7. 7. Mooney MR, Unger BT, Boland LL, et al. Therapeutic hypothermia after out-ofhospital cardiac arrest: evaluation of a regional system to increase access to cooling. Circulation 2011;124:206–14. 8. Sendelbach S, Hearst MO, Johnson PJ, Unger BT, Mooney MR. Effects of variation in temperature management on cerebral performance category scores in patients who received therapeutic hypothermia post cardiac arrest. Resuscitation 2012;83:829–34. 9. Wolff B, Machill K, Schumacher D, Schulzki I, Werner D. Early achievement of mild therapeutic hypothermia and the neurologic outcome after cardiac arrest. Int J Cardiol 2009;133:223–8.
7
10. Benz-Woerner J, Delodder F, Benz R, et al. Body temperature regulation and outcome after cardiac arrest and therapeutic hypothermia. Resuscitation 2012;83:338–42. 11. Haugk M, Testori C, Sterz F, et al. Relationship between time to target temperature and outcome in patients treated with therapeutic hypothermia after cardiac arrest. Crit Care 2011;15:R101. 12. Italian Cooling Experience (ICE) Study Group. Early-versus late-initiation of therapeutic hypothermia after cardiac arrest: preliminary observations from the experience of 17 Italian intensive care units. Resuscitation 2012;83:823–8. 13. Nielsen N, Hovdenes J, Nilsson F, et al. Outcome, timing and adverse events in therapeutic hypothermia after out-of-hospital cardiac arrest. Acta Anaesthesiol Scand 2009;53:926–34. 14. Perman SM, Ellenberg JH, Grossestreuer AV, et al. Shorter time to target temperature is associated with poor neurologic outcome in post-arrest patients treated with targeted temperature management. Resuscitation 2015;88:114–9. 15. Vincent JL, de Mendonc¸a A, Cantraine F, et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on sepsis-related problems of the European Society of Intensive Care Medicine. Crit Care Med 1998;26: 1793–800. 16. Booth CM, Boone RH, Tomlinson G, Detsky AS. Is this patient dead, vegetative, or severely neurologically impaired? Assessing outcome for comatose survivors of cardiac arrest. JAMA 2004;291:870–9. 17. Murnin MR, Sonder P, Janssens GN, et al. Determinants of heat generation in patients treated with therapeutic hypothermia following cardiac arrest. J Am Heart Assoc 2014;3:e000580. 18. Kaneko T, Kasaoka S, Nakahara T, et al. Effectiveness of lower target temperature therapeutic hypothermia in post-cardiac arrest syndrome patients with a resuscitation interval of ≤30 min. J Intensive Care 2015;3:28. 19. Wallmüller C, Testori C, Sterz F, et al. Limited effect of mild therapeutic hypothermia on outcome after prolonged resuscitation. Resuscitation 2016;98:15–9. 20. Coimbra C, Wieloch T. Moderate hypothermia mitigates neuronal damage in the rat brain when initiated several hours following transient cerebral ischemia. Acta Neuropathol 1994;87:325–31. 21. Ohta H, Terao Y, Shintani Y, Kiyota Y. Therapeutic time window of post-ischemic mild hypothermia and the gene expression associated with the neuroprotection in rat focal cerebral ischemia. Neurosci Res 2007;57:424–33. 22. Abella BS, Rhee JW, Huang KN, Vanden Hoek TL, Becker LB. Induced hypothermia is underused after resuscitation from cardiac arrest: a current practice survey. Resuscitation 2005;64:181–6. 23. Patel PV, John S, Garg RK, Temes RE, Bleck TP, Prabhakaran S. Therapeutic hypothermia after cardiac arrest is underutilized in the United States. Ther Hypothermia Temp Manage 2011;1:199–203. 24. Lee SJ, Jeung KW, Lee BK, et al. Impact of case volume on outcome and performance of targeted temperature management in out-of-hospital cardiac arrest survivors. Am J Emerg Med 2015;33:31–6. 25. Komatsu T, Kinoshita K, Sakurai A, et al. Shorter time until return of spontaneous circulation is the only independent factor for a good neurological outcome in patients with postcardiac arrest syndrome. Emerg Med J 2014;31:549–55. 26. Terman SW, Shields TA, Hume B, Silbergleit R. The influence of age and chronic medical conditions on neurological outcomes in out of hospital cardiac arrest. Resuscitation 2015;89:169–76.
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
ARTICLE IN PRESS
RESUS-7001; No. of Pages 7
Resuscitation xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management夽,夽夽 Byung Kook Lee a , Kyung Woon Jeung a,∗ , Yong Hun Jung a , Dong Hun Lee a , Sung Min Lee a , Yong Soo Cho a , Tag Heo a , Jong Geun Yun b , Yong Il Min a a b
Department of Emergency Medicine, Chonnam National University Hospital, 42 Jebong-ro, Donggu, Gwangju, Republic of Korea Department of Emergency Medical Services, Honam University, 417 Eodeung-daero, Gwangsangu, Gwangju, Republic of Korea
a r t i c l e
i n f o
Article history: Received 28 August 2016 Received in revised form 28 November 2016 Accepted 4 December 2016 Keywords: Heart arrest Induced hypothermia Body temperature regulation Prognosis
a b s t r a c t Aim of the study: Studies examining associations between time to target temperature and outcomes in cardiac arrest patients who underwent targeted temperature management (TTM) have shown inconsistent results. We examined these associations separately for time from restoration of spontaneous circulation to TTM initiation (pre-induction time) and time from TTM initiation to target temperature (induction time). Furthermore, we examined whether critical time thresholds exist if there is an association. Methods: This was a single-centre retrospective observational study including adult cardiac arrest patients treated with TTM from 2008 to 2015. We tested the associations of pre-induction time and induction time with outcomes at hospital discharge using multivariate logistic regression analysis. We then performed additional multivariate analyses, each with the significant timing variable at different binary cutoffs. Results: A total of 515 patients were analysed. At hospital discharge, 357 patients (69.3%) were alive, of whom 161 (31.3%) had a favourable neurologic outcome. In multivariate analysis, a shorter preinduction time was independently associated with a favourable neurologic outcome (odds ratio [OR], 1.110; 95% confidence interval [CI], 1.025–1.202), whereas the induction time was not (OR, 0.954; 95% CI, 0.852–1.067). We found two pre-induction time thresholds (120 and 360 min) that were associated with neurologic outcome. Conclusion: We found that a shorter pre-induction time was independently associated with a favorable neurologic outcome at hospital discharge, whereas induction time was not. We also found two time thresholds at 120 and 360 min, after which initiation of cooling was associated with a worse neurologic outcome. © 2016 Published by Elsevier Ireland Ltd.
Introduction
Abbreviations: TTM, targeted temperature management; ROSC, restoration of spontaneous circulation; CNUH, Chonnam national university hospital; ECMO, extracorporeal membrane oxygenation; CPR, cardiopulmonary resuscitation; GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; CPC, Cerebral Performance Categories; OR, odds ratio; CI, confidence interval. 夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at http://dx.doi.org/10.1016/j.resuscitation.2016.12.002. 夽夽 Chonnam National University Hospital Institutional Review Board Protocol No. CNUH-2016-104. ∗ Corresponding author. Fax: +82 62 228 7417. E-mail addresses: [email protected] (B.K. Lee), [email protected] (K.W. Jeung), [email protected] (Y.H. Jung), [email protected] (D.H. Lee), [email protected] (S.M. Lee), [email protected] (Y.S. Cho), [email protected] (T. Heo), [email protected] (J.G. Yun), [email protected] (Y.I. Min).
Targeted temperature management (TTM) has been widely recommended for the treatment of comatose cardiac arrest survivors since the publication of two landmark papers on therapeutic hypothermia.1,2 Despite a number of studies regarding the TTM after cardiac arrest, many questions remain unanswered. Among them, the relationship between timing of cooling and outcome is an unresolved issue that bears important clinical implications. Animal studies have consistently shown that earlier cooling results in greater benefits from this therapy.3–5 However, human studies have shown inconsistent results on the impact of earlier cooling.6–14 It appears reasonable to examine the association between time to target temperature, which is the time interval from restoration
http://dx.doi.org/10.1016/j.resuscitation.2016.12.002 0300-9572/© 2016 Published by Elsevier Ireland Ltd.
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model RESUS-7001; No. of Pages 7
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
2
36–36.5 ◦ C. All patients undergoing TTM received continuous intravenous midazolam and remifentanil (or fentanyl) until at least 72 h after cardiac arrest. A neuromuscular blocking agent was administered to control shivering on an as-needed basis. All other aspects of patient management were at the discretion of the treating physicians.
of spontaneous circulation (ROSC) to achievement of target temperature, and outcomes in order to understand how the timing of cooling impacts outcomes of comatose cardiac arrest survivors. The time to target temperature can be divided into two time intervals: pre-induction time, which is the time interval from ROSC to TTM initiation, and induction time, the time interval from TTM initiation to achievement of target temperature. According to the animal studies showing benefits of earlier cooling, a shorter pre-induction time is expected to be associated with favourable outcomes.3–5 In contrast, clinical studies suggest that a shorter induction time is associated with worse outcomes.11,14 Thus, regarding the preinduction time and induction time, the associations with outcomes can be in the opposite direction, and the direction and significance of the association between time to target temperature and outcomes may vary depending on the relative contribution of the two time intervals to outcomes, thus resulting in inconsistent results in human studies. In light of this, it may be more reasonable to distinguish between pre-induction time and induction time in exploring the association between the timing of cooling and outcomes. To our knowledge, few studies examined this association separately for the preinduction and induction times.11,14 In the present study, we sought to examine the associations of pre-induction time and induction time with survival to hospital discharge and neurologic outcome at hospital discharge. Furthermore, we sought to examine whether critical time thresholds exist if there is an association. We hypothesised that a shorter pre-induction time would be associated with favourable outcomes while a shorter induction time would be associated with unfavourable outcomes.
The following data were obtained from hospital records: age, sex, comorbidities, first monitored rhythm, aetiology of cardiac arrest, location of cardiac arrest, presence of a witness on collapse, bystander cardiopulmonary resuscitation (CPR), time to ROSC, transfer status (initially presented at CNUH versus initially presented at another hospital and transferred to CNUH), Glasgow Coma Scale (GCS) score after ROSC, initial temperature (recorded on hospital admission), TTM device, pre-induction time, induction time, duration of rewarming, sequential organ failure assessment (SOFA) score within the first 24 h after admission,15 vital status at hospital discharge (alive or dead), and neurologic outcome at hospital discharge. For the calculation of pre-induction and induction times, the initiation of TTM was defined as the time when the first attempt of cooling was made. Neurologic outcome was assessed using the Glasgow–Pittsburgh Cerebral Performance Categories (CPC) scale at discharge and recorded as CPC 1 (good performance), CPC 2 (moderate disability), CPC 3 (severe disability), CPC 4 (vegetative state), or CPC 5 (brain death or death).16 The primary outcome was an unfavourable neurological outcome, defined as CPC 3–5 at hospital discharge. The secondary outcome was in-hospital mortality.
Methods
Statistical analysis
Study design and population
Categorical variables were presented as frequencies and percentages. Comparisons of categorical variables were performed using 2 or Fisher exact tests, as appropriate. Continuous variables were presented as median values with interquartile ranges because all continuous variables showed non-normal distribution. The Mann–Whitney U test was conducted for comparisons of continuous variables. The Kruskal–Wallis test was performed to compare induction times among the cooling devices. General linear model analysis was performed to identify the interaction between transfer status and location of cardiac arrest. Multivariate logistic regression analysis was used to examine the association between the timing of cooling and outcomes after adjusting for potential confounders. All variables with p < 0.2 in univariate analyses were included in the multivariate regression model. The multicollinearity between variables was assessed before modelling. Backward selection was used to obtain the final model. First, we entered the timing variables into the model as continuous variables. Second, to examine whether a significant time threshold value could be determined, we performed additional multivariate logistic regression analyses, each with the significant timing variable at different binary cutoffs (60-min increments from 60 to 420 min of preinduction time). Data were analysed using PASW/SPSSTM software, version 18 (IBM Inc., Chicago, IL, USA). A two-sided significance level of 0.05 was used for statistical significance.
This was a retrospective observational cohort study of adult comatose cardiac arrest survivors treated with TTM at Chonnam National University Hospital (CNUH) in Gwangju, Korea, from January 2008 to December 2015. The Institutional Review Board of CNUH approved this study. Cardiac arrest patients over 18 years of age who underwent TTM were included. Patients were excluded if (1) TTM was interrupted owing to transfer to another facility or haemodynamic instability, (2) they died before the target temperature was reached, (3) they were treated with a different TTM protocol, or (4) extracorporeal membrane oxygenation was applied during post-cardiac arrest care. TTM protocol No cooling attempt was made in the pre-hospital phase, and thus TTM was started after arrival at our hospital in all patients of this study. According to our protocol, TTM was considered for all nontraumatic cardiac arrest survivors who could not obey commands. Patients were not eligible for TTM if they had intracranial haemorrhage, active bleeding, known terminal illness, or a poor pre-arrest neurologic status. Cooling was initiated as soon as possible with ice packs, intravenous cold saline, and one of the ® following three TTM devices: Arctic Sun Energy Transfer PadsTM ® (Medivance Corp, Louisville, KY, USA); Blanketrol II (Cincinnati ® Subzero Products, Cincinnati, OH, USA); or COOLGARD3000 Thermal Regulation System (Alsius Corporation, Irvine, CA, USA). The target temperature of 33 ◦ C was achieved and then maintained for 24 h using one of the TTM devices. Upon completion of the 24-h maintenance phase, active rewarming was attempted at a target rate of 0.25 ◦ C–0.5 ◦ C h−1 until the body temperature reached
Data collection
Results A total of 636 adult cardiac arrest patients were treated with TTM during the study period. Of these, 121 patients were excluded as shown in Fig. 1. Thus, 515 patients were included in this study. At hospital discharge, 357 patients (69.3%) were alive, of whom 161 (31.3%) had a favourable neurologic outcome. Clinical characteristics stratified by outcomes are shown in Table 1. Patients
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
Unfavourable neurologic outcome (n = 354)
p-Value
Survivor (n = 357)
Nonsurvivor (n = 158)
p-Value
60.0 (47.0–70.0) 341 (66.2)
52.0 (42.0–61.0) 117 (72.7)
64.0 (51.0–72.0) 224 (63.3)
<0.001 0.037
58.0 (46.0–69.0) 235 (65.8)
65.0 (51.0–73.0) 106 (67.1)
<0.001 0.780
75 (14.6) 39 (7.6) 201 (39.0) 135 (26.2) 30 (5.8) 54 (10.5) 38 (7.4) 12 (2.3) 402 (78.1)
27 (16.8) 77 (6.8) 48 (29.8) 21 (13.0) 3 (1.9) 9 (5.6) 6 (3.7) 2 (1.2) 124 (77.0)
48 (13.6) 28 (7.9) 153 (43.2) 114 (32.2) 27 (7.6) 45 (12.7) 32 (9.0) 10 (2.8) 278 (78.5)
0.338 0.668 0.004 <0.001 0.010 0.014 0.033 0.270 0.701 0.937
49 (13.7) 26 (7.3) 131 (36.7) 80 (22.4) 19 (5.3) 28 (7.8) 27 (7.6) 9 (2.5) 282 (79.0)
26 (16.5) 13 (8.2) 70 (44.3) 55 (34.8) 11 (7.0) 26 (16.5) 11 (7.0) 3 (1.9) 120 (75.9)
0.418 0.709 0.103 0.003 0.464 0.003 0.810 0.666 0.442 0.044
418 (81.2) 97 (18.8)
131 (81.4) 30 (18.6)
287 (81.1) 67 (18.9)
298 (83.5) 59 (16.5)
120 (75.9) 38 (24.1)
153 (29.7) 111 (21.6) 217 (42.1) 34 (6.6)
97 (60.2) 31 (19.3) 24 (14.9) 9 (5.6)
56 (15.8) 80 (22.6) 193 (54.5) 25 (7.1)
129 (36.1) 75 (21.0) 127 (35.6) 26 (7.3)
24 (15.2) 36 (22.8) 90 (57.0) 8 (5.1)
282 (54.8) 117 (22.7) 74 (14.4) 38 (7.4) 4 (0.8) 388 (75.3) 254 (49.3) 28.0 (17.0–40.0) 3 (3–4) 36.0 (35.0–36.7)
132 (82.0) 13 (8.1) 5 (3.1) 11 (6.8) 0 (0.0) 140 (87.0) 89 (55.3) 21.0 (13.0–30.0) 4 (3–6) 36.4 (35.9–37.0)
150 (42.4) 104 (29.4) 69 (19.5) 27 (7.6) 4 (1.1) 248 (70.1) 165 (46.6) 30.0 (20.0–40.0) 3 (3–3) 36.0 (34.7–36.5)
215 (60.2) 68 (19.0) 50 (14.0) 22 (6.2) 2 (0.6) 277 (77.6) 180 (50.4) 25.0 (15.0–36.5) 3 (3–4) 36.0 (35.3–36.8)
67 (42.4) 49 (31.0) 24 (15.2) 16 (10.1) 2 (1.3) 111 (70.3) 74 (46.8) 32.0 (20.0–42.5) 3 (3–3) 35.8 (34.6–36.5)
137 (26.6) 135 (26.2) 243 (47.2) 215 (150–300) 2.5 (1.5–4.0) 12.0 (10.4–13.6), 482c 10 (7–12)
39 (24.2) 48 (29.8) 74 (46.0) 202 (150–300) 3.0 (2.0–5.0) 12.0 (10.0–13.0) 8 (6–10)
98 (27.7) 87 (24.6) 169 (47.7) 225 (155–302) 2.0 (1.2–3.5) 12.0 (11.0–14.0), 321c 11 (8–12)
90 (25.2) 92 (25.8) 175 (49.0) 230 (159–300) 2.8 (1.8–4.0) 12.0 (10.5 − 13.0) 9 (6–11)
47 (29.7) 43 (27.2) 68 (430) 198 (145–300) 2.0 (1.0–3.5) 12.0 (10.0–14.0), 125c 11 (10–14)
<0.001
<0.001
<0.001
<0.001 0.068 <0.001 <0.001 <0.001 0.421
0.105 <0.001 0.029 <0.001
0.003
0.075 0.453 0.001 <0.001 0.001 0.414
0.143 <0.001 0.549 <0.001
CAD, coronary artery disease; CVA, cerebrovascular accident; VF, ventricular fibrillation; VT, ventricular tachycardia; PEA, pulseless electrical activity; CPR, cardiopulmonary resuscitation; ROSC, restoration of spontaneous circulation; SOFA, sequential organ failure assessment. a Patients initially presented to another hospital and then transferred to Chonnam National University Hospital. b Of the 418 out-of-hospital cardiac arrest patients, 90 initially presented at CNUH, whereas 328 initially presented at another hospital and were transferred to CNUH. Among the 97 in-hospital cardiac arrest patients, 23 experienced cardiac arrest in CNUH, whereas 74 experienced cardiac arrest in another hospital and were transferred to CNUH. In general linear model univariate analysis, no interaction was found between transfer status and location of cardiac arrest for either the neurologic outcome (p = 0.407) or in-hospital mortality (p = 0.804). c Number of patients for whom rewarming duration was available.
ARTICLE IN PRESS
Comorbidities CAD Heart failure Hypertension Diabetes Pulmonary disease Renal disease CVA Hepatic disease Transfera Locationb Out-of-hospital In-hospital First monitored rhythm VF/pulseless VT PEA Asystole Unknown Aetiology Cardiac Other medical Asphyxia Drug overdose Drowning Witnessed collapse Bystander CPR Time to ROSC (minutes) Glasgow coma scale Initial temperature (◦ C) Cooling device Blanket Endovascular Hydrogel pad Pre-induction time (minutes) Induction time (hours) Rewarming duration (hours) SOFA score
Favourable neurologic outcome (n = 161)
G Model
RESUS-7001; No. of Pages 7
Age (years) Male sex
Total (n = 515)
B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
Table 1 Demographic and arrest characteristics stratified by outcomes.
3
G Model RESUS-7001; No. of Pages 7 4
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
Fig. 1. Schematic diagram showing the number of patients included in the present study. TTM, targeted temperature management; ECMO, extracorporeal membrane oxygenation.
Fig. 2. Neurologic outcome according to the pre-induction time (A) and the induction time (B).
with favourable neurologic outcome were younger and predominantly male. They had a lower incidence of comorbidities, were more likely to have a shockable rhythm, were more likely to have a cardiac aetiology, had a higher incidence of witnessed collapse, had a shorter time to ROSC, had a higher GCS score after ROSC, had a higher initial temperature, had a shorter duration of rewarming, and had a lower SOFA score compared with patients with
unfavourable neurologic outcome. Fig. 2 shows the neurologic outcome of the patients according to the pre-induction and induction times. The pre-induction time did not differ between favourable and unfavourable neurologic outcome groups, whereas the induction time was longer in patients with favorable neurologic outcome. The pre-induction time was significantly longer in transferred patients (235 [166–312] min) than in nontransferred patients (157
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model
ARTICLE IN PRESS
RESUS-7001; No. of Pages 7
B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
5
Table 2 Multivariate logistic regression analysis for an unfavourable neurologic outcome and in-hospital mortality.
Age (years) Diabetes Pulmonary disease Previous CVA Shockable rhythm Cardiac aetiology Time to ROSC (every 5 min) GCS Initial body temperature (◦ C) SOFA score Pre-induction time (every 30 min) Induction time (hours)
Unfavourable neurologic outcome OR (95% CI)
p-Value
In-hospital mortality OR (95% CI)
p-Value
1.027 (1.008–1.046) 2.051 (1.039–4.167) 7.497 (1.160–48.446) 6.863 (1.835–25.660) 0.222 (0.114–0.435) 0.266 (0.136–0.519) 1.290 (1.163–1.432) 0.546 (0.435–0.686) 0.798 (0.630–1.011) 1.099 (1.004–1.203) 1.110 (1.025–1.202) 0.954 (0.852–1.067)
0.006 0.039 0.034 0.004 <0.001 <0.001 <0.001 <0.001 0.062 0.041 0.010 0.408
1.002 (0.985–1.020) 1.190 (0.660–2.753) NA NA 0.347 (0.188–0.639) 0.800 (0.466–1.374) 1.114 (1.030–1.204) 0.630 (0.475–0.835) 1.092 (0.925–1.289) 1.225 (1.137–1.319) 0.965 (0.905–1.030) 0.949 (0.848–1.062)
0.801 0.413
0.001 0.419 0.007 0.001 0.300 <0.001 0.286 0.360
Pre-induction time and induction time were entered into the model as continuous variables. OR, odds ratio; CI, confidence interval; CVA, cerebrovascular accident; ROSC, restoration of spontaneous circulation; GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; NA, not available.
[115–237] min; p < 0.001). The induction time differed significantly according to the TTM device used (3.3 [2.0–5.4] h for the blanket, 2.0 [1.0–4.0] h for the endovascular cooling catheter, and 2.3 [1.3–3.3] h for the hydrogel pad; p < 0.001). Survivors were younger and had a lower incidence of comorbidities. They were more likely to have a shockable rhythm and cardiac aetiology and were more likely to have had an out-of-hospital cardiac arrest (Table 1). Survivors had a shorter time to ROSC, higher GCS score, higher initial temperature, and lower SOFA score compared with nonsurvivors. Pre-induction time did not differ between survivors and nonsurvivors, whereas induction time was significantly longer in survivors. Association of pre-induction time and induction time with neurologic outcome Table 2 shows the result of multivariate logistic regression analysis with pre-induction time and induction time entered as continuous variables. A shorter pre-induction time was independently associated with a favourable neurologic outcome (odds ratio [OR], 1.110; 95% confidence interval [CI], 1.025–1.202), whereas induction time was no longer significantly associated with the neurologic outcome (OR, 0.954; 95% CI, 0.852–1.067). When the patients were divided into two groups according to the time to ROSC (median 28 min), the association between pre-induction time and neurologic outcome was significant only in patients with shorter time to ROSC (Fig. 3). Additional multivariate analyses performed with the pre-induction time at different binary thresholds revealed threshold effects at the pre-induction times of 120 and 360 min (Table 3). Pre-induction times longer than 120 min (OR, 2.264; 95% CI, 1.026–4.995) and longer than 360 min (OR, 2.694; 95% CI, 1.139–6.372) were independently associated with an unfavourable neurologic outcome compared with pre-induction times of ≤120 min and ≤360 min, respectively (Table 3). Table 3 Multivariate logistic models of association between pre-induction time at different binary cutoffs and unfavourable neurologic outcome. Pre-induction time (min)
OR (95% CI)
p-Value
>60 >120 >180 >240 >300 >360 >420
0.801 (0.039–16.551) 2.264 (1.026–4.995) 1.585 (0.906–2.774) 1.265 (0.716–2.235) 1.319 (0.686–2.537) 2.694 (1.139–6.372) 2.497 (0.961–6.490)
0.886 0.043 0.107 0.419 0.407 0.024 0.060
OR, odds ratio; CI, confidence interval.
Fig. 3. Multivariate logistic regression analysis for poor neurologic outcome in patients with time to restoration of spontaneous circulation ≤28 min (A) and >28 min (B). SOFA, sequential organ failure assessment; GCS, Glasgow coma scale, CI, confidence interval.
Association of pre-induction time and induction time with in-hospital mortality Multivariate analysis showed that shockable rhythm (OR, 0.360; 95% CI, 0.196–0.661), time to ROSC (OR, 1.014; 95% CI, 1.002–1.027), GCS (OR, 0.622; 95% CI, 0.469–0.824), and SOFA score (OR, 1.225; 95% CI, 1.137–1.320) were independently associated with in-hospital mortality. Neither the pre-induction time nor the induction time was associated with in-hospital mortality in the multivariate analysis (Table 2).
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model RESUS-7001; No. of Pages 7
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
6
Discussion In this study, we found that a shorter pre-induction time was independently associated with a favourable neurologic outcome at hospital discharge, whereas induction time had no independent association with neurologic outcome. We also found two preinduction time thresholds (120 and 360 min), after which initiation of cooling was associated with a worse neurologic outcome. Studies have suggested that a more precipitous temperature drop induced by cooling is associated with worse outcomes,11,14 reflecting a loss of the ability to resist cooling owing to impaired thermoregulatory control. In a multicentre observational study including 321 comatose cardiac arrest patients, patients with an induction time >300 min had a higher likelihood of a favourable neurologic outcome compared with those with an induction time <120 min.14 In line with these studies,11,14 greater heat generation, which impedes the cooling process, was associated with reduced ischaemic injury and better neurologic outcome in a study that evaluated patient heat generation during TTM after cardiac arrest.17 In our study, the induction time was longer in patients with a favourable neurologic outcome, although this association did not reach statistical significance in multivariate analysis. Nonetheless, our study, along with the previous studies showing shorter induction time in patients with worse outcomes, suggests that preinduction time and induction time should be analysed separately when exploring the association between timing of cooling and outcomes. Similar to our study, Perman et al. and Haugk et al. distinguished the pre-induction time and induction time in exploring the association between time to target temperature and outcomes.11,14 However, in contrast to our study, pre-induction time did not differ significantly between patients with favourable versus unfavourable neurologic outcomes in their studies. The reasons for the difference between our findings and those of these studies are not readily apparent. One possible explanation is that compared with these studies, the pre-induction time in our study was relatively long and had a wide range, which might be adequate to reveal the threshold effects present at pre-induction time >120 min. Our study included a large proportion of patients transferred from district or secondary hospitals in a large rural area, as well as patients who experienced cardiac arrest at our hospital; this resulted in a long pre-induction time with a wide range. In the present study, the association between pre-induction time and neurologic outcome was significant in patients with time to ROSC ≤28 min, whereas this association was not significant in patients with time to ROSC >28 min. This finding is in line with the results of several studies suggesting loss of efficacy of cooling in patients with prolonged time to ROSC.18,19 In a study that compared neurologic outcome between two target temperature ranges (32.0–33.5 ◦ C versus 34.0–35.0 ◦ C) using a multicentre observational registry of post-cardiac arrest patients treated with TTM, cooling was associated with improved neurologic recovery in patients with time to ROSC ≤30 min, whereas it was not in patients with time to ROSC >30 min.18 To our knowledge, no study has explored whether a time threshold could be determined for the neuroprotection conferred by cooling. In the present study, we found two pre-induction time thresholds at 120 and 360 min. This finding should be interpreted in light of some limitations. We did not find a significant threshold at pre-induction time <120 or >360 min. However, considering the relatively small numbers of patients with pre-induction time <120 or >360 min in our study, further studies are required to determine whether a critical threshold exists at pre-induction time <120 or >360 min. Experimental studies have indicated that hypothermia mitigates neuronal injury even when initiated several hours after
ischaemia.20,21 However, despite a great number of clinical studies on TTM after cardiac arrest, the time limit within which cooling should be initiated to achieve its therapeutic benefit remains unknown. Currently, pre-hospital cooling is not widely adopted owing to a lack of firm evidence supporting its use. Furthermore, TTM is limited mostly to a few large hospitals in many countries.22–24 Thus, many cardiac arrest patients are transferred to a TTM-capable facility with the ensuing loss of valuable time and are often admitted to the TTM-capable facility after several hours have elapsed since the cardiac arrest. Information on the window of opportunity is important, particularly to optimise the care of these patients. Our study did not directly address the window of opportunity for hypothermia. Nonetheless, the presence of a pre-induction time threshold at 360 min suggests that the time window of the initiation of cooling should be at most 6 h after ROSC. In this study, pre-induction time was independently associated with neurologic outcome, but it had no significant association with in-hospital mortality. Although this finding is in line with that of a previous study in which early cooling improved neurologic outcome rather than survival,4 this finding should be interpreted cautiously, because our study did not evaluate long-term survival and patients with poor neurologic outcome at discharge may have a higher risk of long-term mortality. In our study, the presence of a witness on collapse and bystander CPR, both of which have traditionally been regarded as important variables influencing cardiac arrest outcome, were not independently associated with neurologic outcome. The reason for this finding is not readily apparent, but the adjustment of time to ROSC might have weakened the effects of these variables in our multivariate model. Consistent with our finding, Komatsu et al. investigated the association of several prehospital and in-hospital factors, including the presence of witness and bystander CPR with outcome in 227 cardiac arrest survivors, and reported that time to ROSC was the only independent factor related to neurologic outcome.25 On the other hand, comorbidities including pulmonary disease and previous CVA were independently associated with neurologic outcome, with adjusted OR of 7.4 and 7.0, respectively. The effect size of these variables should be assessed through further study, since confidence intervals of these variables were large in our multivariate model. Moreover, a recent study suggests that comorbidity itself is not independently associated with neurologic outcome.26 Our study has several limitations. First, this was a single-centre study. Thus, our results require further confirmation by other studies. Second, generalisability to other country settings is uncertain. In Korea, physicians are generally not permitted to withdraw lifesustaining therapies unless the patient is pronounced brain dead. Thus, treatment was not withdrawn for any patient included in this study. This may explain, at least in part, the high proportion of patients with an unfavourable neurologic outcome in the present study. Third, it was a retrospective observational study, and thus, selection bias could have been present despite our efforts to control for this. Fourth, because of its observational nature, our study identified only association and could not determine a causal relationship. Fifth, several factors that could possibly influence the efficacy of TTM, including number of ice packs used, amount of cold saline infused, body habitus, proportion of body surface area covered by cooling pads, and administration of anti-shivering medications, were not fully investigated in our study. Sixth, despite our efforts, there may have been unmeasured confounding factors. For example, transferred patients might not only experience delay in TTM initiation, but could also have received suboptimal care before admission to our hospital. Furthermore, patients with more severe injury might have been transferred to our hospital for TTM. Seventh, pre-induction time involves systemic issues and a number of
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002
G Model RESUS-7001; No. of Pages 7
ARTICLE IN PRESS B.K. Lee et al. / Resuscitation xxx (2016) xxx–xxx
factors can influence the length of pre-induction time. However, our study did not address this issue. Conclusions In this study, we found that a shorter pre-induction time was independently associated with a favourable neurologic outcome at hospital discharge whereas induction time had no independent association with outcome. We also found two pre-induction time thresholds at 120 and 360 min, after which initiation of cooling was associated with a worse neurologic outcome. Our results suggest that pre-induction time and induction time should be analysed separately when exploring the association between the timing of cooling and outcomes. Conflict of interest statement All authors have no potential conflicts of interest to disclose. References 1. Bernard SA, Gray TW, Buist MD, et al. Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. N Engl J Med 2002;346:557–63. 2. Hypothermia after Cardiac Arrest Study Group. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Engl J Med 2002;346:549–56. 3. Che D, Li L, Kopil CM, Liu Z, Guo W, Neumar RW. Impact of therapeutic hypothermia onset and duration on survival, neurologic function, and neurodegeneration after cardiac arrest. Crit Care Med 2011;39:1423–30. 4. Kuboyama K, Safar P, Radovsky A, Tisherman SA, Stezoski SW, Alexander H. Delay in cooling negates the beneficial effect of mild resuscitative cerebral hypothermia after cardiac arrest in dogs: a prospective, randomized study. Crit Care Med 1993;21:1348–58. 5. Nozari A, Safar P, Stezoski SW, et al. Critical time window for intra-arrest cooling with cold saline flush in a dog model of cardiopulmonary resuscitation. Circulation 2006;113:2690–6. 6. Chiota NA, Freeman WD, Barrett K. Earlier hypothermia attainment is associated with improved outcomes after cardiac arrest. J Vasc Interv Neurol 2011;4:14–7. 7. Mooney MR, Unger BT, Boland LL, et al. Therapeutic hypothermia after out-ofhospital cardiac arrest: evaluation of a regional system to increase access to cooling. Circulation 2011;124:206–14. 8. Sendelbach S, Hearst MO, Johnson PJ, Unger BT, Mooney MR. Effects of variation in temperature management on cerebral performance category scores in patients who received therapeutic hypothermia post cardiac arrest. Resuscitation 2012;83:829–34. 9. Wolff B, Machill K, Schumacher D, Schulzki I, Werner D. Early achievement of mild therapeutic hypothermia and the neurologic outcome after cardiac arrest. Int J Cardiol 2009;133:223–8.
7
10. Benz-Woerner J, Delodder F, Benz R, et al. Body temperature regulation and outcome after cardiac arrest and therapeutic hypothermia. Resuscitation 2012;83:338–42. 11. Haugk M, Testori C, Sterz F, et al. Relationship between time to target temperature and outcome in patients treated with therapeutic hypothermia after cardiac arrest. Crit Care 2011;15:R101. 12. Italian Cooling Experience (ICE) Study Group. Early-versus late-initiation of therapeutic hypothermia after cardiac arrest: preliminary observations from the experience of 17 Italian intensive care units. Resuscitation 2012;83:823–8. 13. Nielsen N, Hovdenes J, Nilsson F, et al. Outcome, timing and adverse events in therapeutic hypothermia after out-of-hospital cardiac arrest. Acta Anaesthesiol Scand 2009;53:926–34. 14. Perman SM, Ellenberg JH, Grossestreuer AV, et al. Shorter time to target temperature is associated with poor neurologic outcome in post-arrest patients treated with targeted temperature management. Resuscitation 2015;88:114–9. 15. Vincent JL, de Mendonc¸a A, Cantraine F, et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on sepsis-related problems of the European Society of Intensive Care Medicine. Crit Care Med 1998;26: 1793–800. 16. Booth CM, Boone RH, Tomlinson G, Detsky AS. Is this patient dead, vegetative, or severely neurologically impaired? Assessing outcome for comatose survivors of cardiac arrest. JAMA 2004;291:870–9. 17. Murnin MR, Sonder P, Janssens GN, et al. Determinants of heat generation in patients treated with therapeutic hypothermia following cardiac arrest. J Am Heart Assoc 2014;3:e000580. 18. Kaneko T, Kasaoka S, Nakahara T, et al. Effectiveness of lower target temperature therapeutic hypothermia in post-cardiac arrest syndrome patients with a resuscitation interval of ≤30 min. J Intensive Care 2015;3:28. 19. Wallmüller C, Testori C, Sterz F, et al. Limited effect of mild therapeutic hypothermia on outcome after prolonged resuscitation. Resuscitation 2016;98:15–9. 20. Coimbra C, Wieloch T. Moderate hypothermia mitigates neuronal damage in the rat brain when initiated several hours following transient cerebral ischemia. Acta Neuropathol 1994;87:325–31. 21. Ohta H, Terao Y, Shintani Y, Kiyota Y. Therapeutic time window of post-ischemic mild hypothermia and the gene expression associated with the neuroprotection in rat focal cerebral ischemia. Neurosci Res 2007;57:424–33. 22. Abella BS, Rhee JW, Huang KN, Vanden Hoek TL, Becker LB. Induced hypothermia is underused after resuscitation from cardiac arrest: a current practice survey. Resuscitation 2005;64:181–6. 23. Patel PV, John S, Garg RK, Temes RE, Bleck TP, Prabhakaran S. Therapeutic hypothermia after cardiac arrest is underutilized in the United States. Ther Hypothermia Temp Manage 2011;1:199–203. 24. Lee SJ, Jeung KW, Lee BK, et al. Impact of case volume on outcome and performance of targeted temperature management in out-of-hospital cardiac arrest survivors. Am J Emerg Med 2015;33:31–6. 25. Komatsu T, Kinoshita K, Sakurai A, et al. Shorter time until return of spontaneous circulation is the only independent factor for a good neurological outcome in patients with postcardiac arrest syndrome. Emerg Med J 2014;31:549–55. 26. Terman SW, Shields TA, Hume B, Silbergleit R. The influence of age and chronic medical conditions on neurological outcomes in out of hospital cardiac arrest. Resuscitation 2015;89:169–76.
Please cite this article in press as: Lee BK, et al. Relationship between timing of cooling and outcomes in adult comatose cardiac arrest patients treated with targeted temperature management. Resuscitation (2016), http://dx.doi.org/10.1016/j.resuscitation.2016.12.002