- Email: [email protected]
Associations between intra-arrest blood glucose level and outcomes of adult in-hospital cardiac arrest: A 10-year retrospective cohort study
Journal Pre-proof Associations between intra-arrest blood glucose level and outcomes of adult in-hospital cardiac arrest: A 10-year retrospective cohort study Chih-Hung Wang, Wei-Tien Chang, Chien-Hua Huang, Min-Shan Tsai, Eric Chou, Ping-Hsun Yu, Yen-Wen Wu, Wen-Jone Chen
PII:
S0300-9572(19)30702-6
DOI:
https://doi.org/10.1016/j.resuscitation.2019.11.012
Reference:
RESUS 8300
To appear in:
Resuscitation
Received Date:
2 August 2019
Revised Date:
14 October 2019
Accepted Date:
16 November 2019
Please cite this article as: Wang C-Hung, Chang W-Tien, Huang C-Hua, Tsai M-Shan, Chou E, Yu P-Hsun, Wu Y-Wen, Chen W-Jone, Associations between intra-arrest blood glucose level and outcomes of adult in-hospital cardiac arrest: A 10-year retrospective cohort study, Resuscitation (2019), doi: https://doi.org/10.1016/j.resuscitation.2019.11.012
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.
1
Associations between intra-arrest blood glucose level and outcomes of adult inhospital cardiac arrest: A 10-year retrospective cohort study
Chih-Hung Wang, MD, PhD1,2; Wei-Tien Chang, MD, PhD1,2; Chien-Hua Huang, MD, PhD1,2; Min-Shan Tsai, MD, PhD1,2; Eric Chou, MD3; Ping-Hsun Yu, MD4; Yen-Wen
Department of Emergency Medicine, National Taiwan University Hospital, Taipei,
-p
1
ro of
Wu, MD, PhD5-7; Wen-Jone Chen, MD, PhD1,2,8
Taiwan
Department of Emergency Medicine, College of Medicine, National Taiwan
re
2
Department of Emergency Medicine, Baylor Scott&White All Saints Medical Center,
na
3
lP
University, Taipei, Taiwan
Fort Worth, Texas, USA
Department of Emergency Medicine, Taipei Hospital, Ministry of Health and Welfare,
ur
4
Jo
New Taipei City, Taiwan 5
Departments of Internal Medicine and Nuclear Medicine, National Taiwan University
Hospital and National Taiwan University College of Medicine, Taipei, Taiwan 6
Department of Nuclear Medicine and Cardiology Division of Cardiovascular Medical
Center, Far Eastern Memorial Hospital, New Taipei City, Taiwan 1
2
7
National Yang-Ming University School of Medicine, Taipei, Taiwan
8
Division of Cardiology, Department of Internal Medicine, National Taiwan University
Hospital and National Taiwan University College of Medicine, Taipei, Taiwan
Address for correspondence and reprints:
ro of
Wen-Jone Chen, MD, PhD No.7, Zhongshan S. Rd., Zhongzheng Dist., Taipei City 100, Taiwan (R.O.C.)
-p
Tel.:+886-2-23123456 ext. 65792
E-mail: [email protected]
re
Fax: +886-2-2322-3150
ur
Abstract
na
the results for publication
lP
The corresponding author had full access to all data and final responsibility to submit
Jo
Aim: We attempted to examine the association between intra-arrest blood glucose (BG) level and outcomes of in-hospital cardiac arrest (IHCA). The interaction between diabetes mellitus (DM) and BG level as well as between dextrose administration and BG level were investigated.
2
3
Methods: This single-centred retrospective study reviewed IHCA patients between 2006 and 2015. Patients with measured intra-arrest BG levels were included. Multivariable logistic regression analyses were conducted. Generalised additive models were used to identify appropriate cut-off points for continuous variables. Interactions
ro of
between independent variables were assessed during the model-fitting process.
Results: Among the 580 included patients, 34 (5.9%) achieved neurologically intact
-p
survival. There were 197 DM patients (34.0%). The mean intra-arrest BG level was
191.5 mg/dl, with 57 patients (9.8%) experiencing hypoglycaemia (BG level≤70 mg/dl).
re
A total of 165 patients (28.4%) received a dextrose injection. An intra-arrest BG
lP
level≤150 mg/dl was inversely associated with favourable neurological outcomes at hospital discharge (odds ratio [OR]: 0.28, 95% confidence interval [CI]: 0.11-0.73; p-
na
value = 0.01). In analyses of interactions, non-DM×BG level≤168 mg/dl was inversely
ur
associated with favourable neurological outcomes (OR: 0.30, 95% CI: 0.11-0.80; p-
Jo
value = 0.02). There were no significant interactions between BG level and dextrose administration.
Conclusion: IHCA patients with intra-arrest BG level≤150 mg/dl had worse neurological recovery. Intra-arrest hypoglycaemia might be a marker of critical illness. 3
4
Dextrose administration was not shown to improve outcomes of IHCA patients with intra-arrest BG level≤150 mg/dl, indicating the need to develop new therapeutics other than dextrose administration for these patients.
Keywords: Glucose; Hypoglycaemia; Diabetes mellitus; In-Hospital cardiac arrest;
lP
re
-p
ro of
Survival; Neurological outcome.
na
Introduction
ur
In the United States, approximately 209,000 patients experience in-hospital cardiac arrest (IHCA) each year [1]. Despite continuing efforts to improve the “chain
Jo
of survival”, IHCA outcomes remain poor. Approximately 24% of IHCA patients survive to hospital discharge; among these patients, about 14% experience significant neurological disability [1]. The 2005 American Heart Association Guidelines for Cardiopulmonary
4
5
Resuscitation and Emergency Cardiovascular Care [2] suggested the providers should identify a factor that may have caused the arrest during cardiopulmonary resuscitation (CPR) and treat it accordingly. Hypoglycaemia had been listed as one of the reversible causes of cardiac arrest in 2005 guidelines [2], but was removed in the 2010 [3] and 2015 [4] guidelines. Whether blood glucose (BG) levels should be measured during
ro of
CPR and whether intra-arrest hypoglycaemia should be treated were not reviewed in the latest 2015 guidelines [4].
-p
Using Get With The Guidelines-Resuscitation (GWTG-R) database, Moskowitz et al. [5] demonstrated that the proportion of IHCA patients receiving dextrose during
re
CPR has increased from 2001 to 2016 despite that the guidelines [3, 4] had removed
lP
the recommendations regarding dextrose administration. As compared to 2001, patients
na
in 2016 had nearly three-fold higher odds of receiving dextrose during CPR [5]. Also using GWTG-R database, Peng et al.[6] indicated that IHCA patients receiving
ur
dextrose administration during CPR had worse neurological and survival outcomes.
Jo
In the studies by Moskowitz et al.[5] and by Peng et al.[6], the actual BG level during CPR was not measured and consequently its confounding effect was not adjusted in the analysis. Hypoglycaemia had been shown to be associated with increased mortality in critically ill patients [7]. Without accounting for BG level in the analysis, the association between dextrose administration and IHCA outcomes may be 5
6
overestimated. Since dextrose has been used increasingly during CPR [5], there is an urgent need to verify its potential harmful effects. Previous studies [8, 9] indicated that optimal BG levels for post-cardiac arrest patients may differ according to the presence of diabetes mellitus (DM). Therefore, in current studies, we attempted not only to elucidate the association between intra-arrest
ro of
BG level and IHCA outcomes, but also the interactions between DM and BG level or
-p
between dextrose administration and BG level.
Materials and Methods
re
Setting
lP
This retrospective cohort study was performed in the tertiary medical centre,
na
National Taiwan University Hospital (NTUH). NTUH has 2,600 beds, including 220 beds in intensive care units (ICUs). This study was conducted in accordance with the
ur
Declaration of Helsinki amendments. The Research Ethics Committee of the NTUH
Jo
approved this study (reference number: 201805098RINC) and waived the requirement for informed consent before data collection. According to hospital policy, a resuscitation team is activated when a cardiac arrest event occurs in the general wards. A resuscitation team consists of a senior resident, several junior residents, a respiratory therapist, a head nurse, and several ICU nurses. Each code team member is certified to 6
7
provide advanced cardiac life support and capable of offering CPR according to current resuscitation guidelines. For cardiac arrest events in the ICUs, a code team is not mobilised since a sufficient number of experienced staffs is always present in the ICUs. In this case, resuscitation is performed by the staff of the ICU where the cardiac arrest
ro of
event occurred and by staff from neighbouring ICUs.
Participants
-p
Patients who experienced IHCA at NTUH from 2006 to 2015 were screened. Patients who met the following criteria were included in the study: (1) aged 18 years or
re
older, (2) documented absence of pulse with performance of chest compressions for at
lP
least 2 minutes, (3) no documentation of a do-not-resuscitate order before arrest, and (4) measurement of BG level during CPR. Sustained return of spontaneous circulation
na
(ROSC) was defined as ROSC lasting ≥ 20 minutes without resumption of chest
ur
compressions. If multiple cardiac arrest events occurred in a single patient during
Jo
hospitalization, only the first event was recorded. Patients who experienced cardiac arrest related to major trauma were excluded from the study.
Data Collection and Outcome Measures The following information was recorded for each patient: age, gender, 7
8
comorbidities [10], variables derived from the Utstein template [11], the first intraarrest BG level measured during CPR, the amount of dextrose administered during CPR, and critical interventions implemented at the time of cardiac arrest and after sustained ROSC. Duration of CPR was defined as the time interval from the first chest compression initiated by the resuscitation team or ICU members to the termination of
ro of
resuscitation efforts, either due to sustained ROSC or declaration of death. Intra-arrest BG level was measured by blood gas analysers. Hypoglycaemia was defined as a BG
-p
level lower than 70 mg/dL (3.9 mmol/L) [12].
The primary outcome was favourable neurological status at hospital discharge.
re
Favourable neurological status was defined as a score of 1 or 2 on the Cerebral
lP
Performance Category (CPC) scale [13]. Patients with a CPC score of 1 or 2 had
na
sufficient cerebral function to live independently. The CPC score was retrospectively determined by reviewing medical records for each patient. The secondary outcome was
Jo
ur
survival to hospital discharge.
Statistical Analysis R 3.3.1 software (R Foundation for Statistical Computing, Vienna, Austria) was
used for data analysis. Categorical data were expressed as counts and proportions; continuous data were expressed as means and standard deviations. Categorical 8
9
variables were compared using the Fisher’s exact test, and continuous variables were examined by the Wilcoxon rank-sum test. A two-tailed p-value of < 0.05 was considered statistically significant. The odds ratio (OR) was selected as the outcome measure, and multivariable logistic regression analyses were performed to examine the associations between the
ro of
independent variables and outcomes. All available independent variables were considered in the regression model, regardless of whether they were significant by
-p
univariate analysis. The stepwise variable selection procedure (with iterations between
the forward and backward steps) was applied to obtain the final regression model.
re
Significance levels for entry and stay were set at 0.15 to avoid exclusion of potential
lP
candidate variables. The final regression model was identified by sequentially
na
excluding individual variables with a p-value > 0.05 until all regression coefficients were statistically significant.
ur
We used generalised additive models (GAMs) [14] to examine the nonlinear
Jo
effects of continuous variables and, if necessary, to identify the appropriate cut-off point(s) for dichotomising a continuous variable during the variable selection procedure. The interaction between DM and BG level or between dextrose administration and BG level was assessed during the model-fitting process. We assessed the goodness-of-fit of the fitted regression model using c statistics, adjusted generalised R2, and the Hosmer9
10
Lemeshow goodness-of-fit test.
Results A total of 1,698 adult non-trauma patients at NTUH received CPR for ≥ 2 min between 2006 and 2015. Of these, 1,118 patients were excluded because of a lack of
ro of
measurement of BG level during CPR. The remaining 580 patients were included for analysis.
-p
Tables 1 and 2 provide the features of cardiac arrest events before, during, and after CPR for all patients in the cohort. The mean age of the patients was 65.6 years.
re
There were 197 diabetic patients (34.0%). A total of 241 cardiac arrest events (41.6%)
lP
occurred in the ICUs, and 292 events (50.3%) occurred on the general wards. The
na
average CPR duration was 40.6 min. The mean intra-arrest BG level was 191.5 mg/dl, with 57 patients (9.8%) having hypoglycaemia. A total of 165 patients (28.4%) received
ur
injections of 50% dextrose in water, with an average of 8.7 g dextrose being
Jo
administered. Only 64 patients (11.0%) survived to hospital discharge; of these, 34 patients (5.9%) demonstrated favourable neurological status. No patients with hypoglycaemia achieved neurologically intact survival in our cohort. We placed all independent variables listed in Tables 1 and 2 in the regression analysis for variable selection. The GAM plots demonstrated a non-linear association 10
11
of logit (p), where p represented the probability for favourable neurological status (Figure 1) or survival (Supplemental Figure 1), with BG level. If logit (p) was greater than zero, the odds for either outcome was greater than one. Therefore, BG level was transformed into a binary variable during the model-fitting process according to the identified cut-off points.
ro of
As shown in Tables 3 and 4, a BG level≤150 mg/dl was inversely associated with favourable neurological outcome (primary model, OR: 0.28, 95% confidence interval
-p
[CI]: 0.11-0.73; p-value = 0.01) and a BG level≤162 mg/dl was inversely associated
with survival (secondary model, OR: 0.39, 95% CI: 0.20-0.76; p-value = 0.006). The
re
analyses of interactions indicated that non-DM×BG level≤168 mg/dl was inversely
lP
associated with the favourable neurological outcome (primary model with interaction
na
term, OR: 0.30, 95% CI: 0.11-0.80; p-value = 0.02; Table 3) and non-DM×BG level≤155 mg/dl was inversely associated with survival (secondary model with
ur
interaction term, OR: 0.28, 95% CI: 0.13-0.59; p-value <0.001; Table 4). There were
Jo
no significant interactions between dextrose administration and BG levels.
Discussion
Main Findings In univariate analysis, we noted that intra-arrest hypoglycaemia (BG level≤70 11
12
mg/dL) was a significant prognosticator of poor neurological outcome. No patients with hypoglycaemia achieved neurologically intact survival in our cohort. In multivariate analysis, a relatively low intra-arrest BG level (BG level≤150 mg/dl), identified by GAM plot (Figure 1), was found to be associated with worse neurological outcomes. When the confounding effects of BG levels were adjusted, dextrose administration was
ro of
not associated with IHCA outcomes, even for patients with relatively low BG levels. In contrast, for non-DM patients, relatively low BG levels were more significantly
re
Comparison with Previous Studies
-p
associated with worse outcomes, in comparison with DM patients.
lP
The practice of empirical dextrose administration in patients with suspected
na
hypoglycaemia-related events, such as altered mental status or cardiac arrest, was based not only on the concerns that delayed treatment may lead to irreversible brain damage,
ur
but also on the unfounded assumption that the dextrose administration was harmless to
Jo
patients who were normoglycemic or hyperglycaemic [15]. Empirical administration of 25 g dextrose (50 mL of 50% dextrose) without waiting for laboratory confirmation of hypoglycaemia has long been recommended for any patient presenting with coma of undetermined aetiology [15]. Nonetheless, there was substantial evidence questioning this practice [16]. Dextrose administration prior to complete cerebral ischemia or during 12
13
severe incomplete ischemia in experimental animals worsened histologic and neurological outcomes [16]. In a randomised-controlled trial, Longstreth et al. [17] randomised 748 patients with out-of-hospital cardiac arrest into two groups according to the intravenous maintenance fluids, either 5% dextrose in water or normal saline. The results indicated neurological outcomes of patients with out-of-hospital cardiac
ro of
arrest did not differ between the two groups, which suggested that dextrose may still be administered to patients during CPR [17]. Nonetheless, the amount of dextrose given
of administering 25 g dextrose during CPR.
-p
in the trial by Longstreth et al. [17] may be too small to be generalized to the practice
re
By using the large GWTG-R registry data, Peng et al.[6] demonstrated that
lP
dextrose was administered to 4173 (4.2%) IHCA patients. Both multivariate regression
na
analysis and propensity-matched analysis indicated that dextrose administration was inversely associated with favourable neurological and survival outcomes [6].
ur
Nevertheless, BG level was not accounted for in the study by Peng et al.[6]. Also
Jo
analysing the data from GWTG-R, Beiser et al.[9] indicated that non-DM patients were sensitive to hypoglycaemia (BG level<70 mg/dl) during the early post-ROSC periods. For other critically ill patients, hypoglycaemia (BG level<70 mg/dl) has also been associated with mortality in a dose-dependent manner, i.e. the lower the BG level, the higher the mortality was [7]. In our cohort, we also noted that for patients with intra13
14
arrest hypoglycaemia (BG level≤70 mg/dl), none regained favourable neurological status. Since hypoglycaemia was one of the indications for administering dextrose [15], without controlling the effects of hypoglycaemia or BG levels, the effects of dextrose administration may be biased by confounding by indications, leading to overestimated associations.
ro of
As demonstrated in our analysis, when the BG level was accounted for in the multivariate analysis, neither dextrose administration nor the amount of dextrose given
-p
was associated with IHCA outcomes. Categorizing a continuous variable without knowing the relationship between independent and dependent variables may lead to
re
loss of information and statistical power [18]. Therefore, we used GAM plots (Figure
lP
1 & Supplemental Figure 1) to identify the optimal cut-off points for BG levels and put
na
the transformed BG levels into multivariate regression analyses for variable selection along with the variable of intra-arrest hypoglycaemia. The results showed intra-arrest
ur
BG levels were significantly associated with outcomes, especially when the BG level
Jo
was relatively low. Interestingly, the upper limit of the relatively low BG level, 150 mg/dl, was over twice the threshold defining hypoglycaemia (70 mg/dl).
Prognostic Role of Intra-arrest BG level In the 2005 resuscitation guidelines [2], hypoglycaemia was listed as one of the 14
15
treatable or reversible causes of cardiac arrest and thus, hypoglycaemia should be identified and treated promptly. Nonetheless, how hypoglycaemia leads to cardiac arrest and whether dextrose administration could reverse this pathophysiological process had not been clearly elucidated in previous studies. In a rat study, Reno et al. [19] demonstrated that severe hypoglycaemia (10-15 mg/dl) induced various cardiac
ro of
arrhythmias, such as premature ventricular contractions, tachycardia, and high-degree heart block. When beta-adrenergic blockers were administered, cardiac arrhythmias
-p
were reduced and hypoglycaemia-related deaths were completely prevented, which
suggested that an overactive sympathoadrenal response may partly account for death
re
related to hypoglycaemia [19]. Therefore, it seemed that dextrose administration alone
na
analyses of interactions.
lP
may not reverse this hypothesized pathophysiological process, as demonstrated in our
Alternatively, hypoglycaemia may just be an epiphenomenon of critical illness. In
ur
a post hoc analysis of the NICE-SUGAR trial [7], investigators revealed that the
Jo
association between hypoglycaemia and mortality was attenuated after adjusting the confounding effects of baseline characteristics and post-randomization factors. Interestingly, the investigators of the NICE-SUGAR trial [7] also reported that for patients not being treated with insulin, the association between hypoglycaemia and mortality was greater, and the time to death was shorter, compared with those receiving 15
16
insulin. The presence of so-called spontaneous hypoglycaemia, i.e. hypoglycaemia not caused by insulin administration, and its strong association with poor outcome highly implied that hypoglycaemia may just be a disease marker and consequently, the hypoglycaemia-targeted therapeutics may not be successful [7]. In our analyses of interactions, we also noted that non-DM patients were more sensitive to relatively low
ro of
BG levels compared with DM patients. Therefore, hypoglycaemia, or relatively low BG levels, may not be a treatable cause; instead, these test results could be viewed as
-p
prognosticators, indicating illness severity. Treatment should thus be focused on
searching for other potential causes leading to cardiac arrest, rather than supplementing
lP
na
Future Applications
re
with dextrose only.
Transient hyperglycaemia during critical illness in patients without DM had been
ur
assumed to be harmless or even advantageous [20]. The development of stress
Jo
hyperglycaemia was thought to be the result of a highly complicated interplay of counter-regulatory hormones, such as catecholamines and cortisol [21, 22]. The absence of stress hyperglycaemia was thought to be caused by a failure in activation of the hypothalamic-pituitary-adrenal axis [20]. As revealed by the rat study of Reno et al. [17], plasma epinephrine levels increased shortly after the onset of severe 16
17
hypoglycaemia with early peaking and then decreasing, possibly indicating adrenal exhaustion. Several studies [23] have also shown that relative adrenal insufficiency was common during the post-ROSC period. Therefore, relatively low intra-arrest BG levels may be an early manifestation of unrecognized relative adrenal insufficiency. For suspected adrenal insufficiency, glucocorticoid administration may be
ro of
considered to improve the haemodynamic response. In a randomized controlled trial by Mentzelopoulos et al.[24] for IHCA, patients allocated to the experimental group
-p
received vasopressin and methylprednisolone during CPR and hydrocortisone for postROSC shock; these patients experienced significantly better outcomes than those in the
re
control group. Nonetheless, in the study by Mentzelopoulos et al.[24], the intra-arrest
lP
BG levels of the two groups did not differ significantly; therefore, whether
na
methylprednisolone supplementation during CPR can improve outcomes for patients with relatively low intra-arrest BG levels should be further examined. For post-ROSC
ur
patients, an evaluation for relative adrenal insufficiency was not frequently performed
Jo
even if they were in shock status [25]. Relatively low intra-arrest BG levels might serve as a sign of relative adrenal insufficiency, reminding clinicians that diagnostic procedures or empirical treatments for relative adrenal insufficiency should be performed.
17
18
Study Limitations First, our study was based on a highly selected cohort, including only 34.2% (580/1698) of all screened patients. Most patients were excluded because of a lack of intra-arrest BG measurement, probably caused by removal of hypoglycaemia from the list of reversible causes of cardiac arrest in resuscitation guidelines [4]. Second, this
ro of
was an observational study, which can only establish an association, rather than a causal relationship, between independent and dependent variables. The exact mechanisms
-p
underlying the association between relatively low intra-arrest BG levels and less favourable outcomes could not be examined in the current analysis. Although relative
re
adrenal insufficiency was suspected to account for this association, the levels of ACTH
lP
and cortisol were not measured in most included patients, as reported in previous
na
studies [25]. Third, the effects of unmeasured confounders may have biased the results despite having used multivariate analysis to adjust the effects of measured confounding
ur
factors. For example, pre-arrest events leading to the relatively low BG levels, pre-
Jo
arrest medication status regarding diabetic control, and the reasons why intra-arrest BG levels were measured may be important confounding factors and could not be verified in a retrospective manner. These limitations can only be solved by a prospective study. Our results may be deemed as hypothesis-generating and serve as the foundation for future study. For example, factors listed in Tables 3 and 4 may be used to design the 18
19
data collection form in the prospective study.
Conclusions Relatively low intra-arrest BG levels (BG≤150 mg/dl) were inversely associated with favourable neurological status at hospital discharge. Dextrose administration was
ro of
not shown to improve outcomes of IHCA patients with relatively low BG levels. For non-DM patients, relatively low BG levels were more significantly associated with
-p
worse outcomes, in comparison with DM patients. These results implied that relatively low intra-arrest BG levels may simply be a disease marker; thus, newer therapeutics
na
lP
re
other than dextrose administration should be developed.
Conflicts of Interests
Jo
ur
The authors declare that they have no conflict of interest.
Acknowledgments We thank Centre of Quality Management of National Taiwan University Hospital for providing the list of patients sustaining in-hospital cardiac arrest. We thank the staff of the 3rd Core Lab, Department of Medical Research, National Taiwan University 19
20
Hospital for technical support. Author Chih-Hung Wang recieved a grant (108-S4091) from the National Taiwan University Hospital. Author Wen-Jone Chen recieved a grant (108-S4236) from the National Taiwan University Hospital. National Taiwan University Hospital had no involvement in designing the study, collecting, analysing or interpreting the data, writing the manuscript, or deciding whether to submit the
Jo
ur
na
lP
re
-p
ro of
manuscript for publication.
20
21
References 1.
Benjamin EJ, Blaha MJ, Chiuve SE, Cushman M, Das SR, Deo R, et al. Heart Disease and Stroke Statistics-2017 Update: A Report From the American Heart Association. Circulation 2017; 135:e146-e603. Part 7.2: Management of Cardiac Arrest. Circulation 2005; 112:IV-58-IV-66.
3.
Neumar RW, Otto CW, Link MS, Kronick SL, Shuster M, Callaway CW, et al.
ro of
2.
Part 8: adult advanced cardiovascular life support: 2010 American Heart
-p
Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2010; 122:S729-67.
Link MS, Berkow LC, Kudenchuk PJ, Halperin HR, Hess EP, Moitra VK, et al.
re
4.
lP
Part 7: Adult Advanced Cardiovascular Life Support: 2015 American Heart
na
Association Guidelines Update for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2015; 132:S444-64. Moskowitz A, Ross CE, Andersen LW, Grossestreuer AV, Berg KM, Donnino
ur
5.
Jo
MW, et al. Trends Over Time in Drug Administration During Adult In-Hospital Cardiac Arrest. Crit Care Med 2019; 47:194-200.
6.
Peng TJ, Andersen LW, Saindon BZ, Giberson TA, Kim WY, Berg K, et al. The administration of dextrose during in-hospital cardiac arrest is associated with increased mortality and neurologic morbidity. Crit Care 2015; 19:160. 21
22
7.
NICE-SUGAR Study Investigators, Finfer S, Liu B, Chittock DR, Norton R, Myburgh JA, et al. Hypoglycemia and risk of death in critically ill patients. N Engl J Med 2012: 367:1108-18.
8.
Wang CH, Huang CH, Chang WT, Tsai MS, Yu PH, Wu YW, et al. Associations between blood glucose level and outcomes of adult in-hospital
9.
ro of
cardiac arrest: a retrospective cohort study. Cardiovasc Diabetol 2016; 15:118. Beiser DG, Carr GE, Edelson DP, Peberdy MA, Hoek TL. Derangements in
-p
blood glucose following initial resuscitation from in-hospital cardiac arrest: a
Resuscitation 2009; 80:624-30.
Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of
lP
10.
re
report from the national registry of cardiopulmonary resuscitation.
na
classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40: 373-83. Jacobs I, Nadkarni V, Bahr J, Berg RA, Billi JE, Bossaert L, et al. Cardiac arrest
ur
11.
Jo
and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries: a statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart 22
23
and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation 2004; 110:3385-97. 12.
Workgroup on Hypoglycemia, American Diabetes Association. Defining and reporting hypoglycemia in diabetes: a report from the American Diabetes Association Workgroup on Hypoglycemia. Diabetes Care 2005; 28:1245-9. Becker LB, Aufderheide TP, Geocadin RG, Callaway CW, Lazar RM, Donnino
ro of
13.
MW, et al. Primary outcomes for resuscitation science studies: a consensus
-p
statement from the American Heart Association. Circulation 2011; 124:215877.
Hastie TJ, Tibshirani RJ. Generalized Additive Models. London and New York:
lP
Chapman & Hall; 1990.
re
14.
Posner JB. The comatose patient. JAMA 1975; 233:1313-4.
16.
Browning RG, Olson DW, Stueven HA, Mateer JR. 50% dextrose: antidote or
na
15.
Longstreth WT Jr, Copass MK, Dennis LK, Rauch-Matthews ME, Stark MS,
Jo
17.
ur
toxin? Ann Emerg Med 1990; 19:683-7.
Cobb LA. Intravenous glucose after out-of-hospital cardiopulmonary arrest: a community-based randomized trial. Neurology 1993; 43:2534-41.
18.
Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ 2006; 332:1080. 23
24
19.
Reno CM, Daphna-Iken D, Chen YS, VanderWeele J, Jethi K, Fisher SJ. Severe hypoglycemia-induced
lethal
cardiac
arrhythmias
are
mediated
by
sympathoadrenal activation. Diabetes 2013; 62:3570-81. 20.
Dungan KM, Braithwaite SS, Preiser JC. Stress hyperglycaemia. Lancet 2009; 373:1798-807. Barth E, Albuszies G, Baumgart K, Matejovic M, Wachter U, Vogt J, et al.
ro of
21.
Glucose metabolism and catecholamines. Crit Care Med 2007; 35:S508-S18.
Andrews RC, Walker BR. Glucocorticoids and insulin resistance: old hormones,
-p
22.
new targets. Clin Sci (Lond) 1999; 96:513-23.
Pene F, Hyvernat H, Mallet V, Cariou A, Carli P, Spaulding C, et al. Prognostic
re
23.
lP
value of relative adrenal insufficiency after out-of-hospital cardiac arrest.
24.
na
Intensive Care Med 2005; 31:627-33.
Mentzelopoulos SD, Malachias S, Chamos C, Konstantopoulos D, Ntaidou T,
ur
Papastylianou A, et al. Vasopressin, steroids, and epinephrine and
Jo
neurologically favorable survival after in-hospital cardiac arrest: a randomized clinical trial. JAMA 2013; 310:270-9.
25.
Miller JB, Donnino MW, Rogan M, Goyal N. Relative adrenal insufficiency in post-cardiac arrest shock is under-recognized. Resuscitation 2008; 76:221-5.
24
25
Figure Legends Figure 1. Generalised additive model plot for nonparametric modelling of the effect of intra-arrest blood glucose level on the logit of probability for favourable neurological
Jo
ur
na
lP
re
-p
ro of
outcomes at hospital discharge
25
26
Table 1. Baseline characteristics of study patients Variables All patients Patients with (n = 580) favourable neurological outcome at hospital discharge (n = 34)
Patients without favourable neurological outcome at hospital discharge (n = 546)
pvalue
Age, y (SDa)
65.6 (16.1)
64.0 (13.0)
65.7 (16.3)
0.33
Male, n (%)
330 (56.9)
23 (67.6)
307 (56.2)
0.22
131 (22.6)
9 (26.5)
105 (18.1)
7 (20.6)
67 (11.6)
7 (20.6)
Heart failure, this admission Heart failure, prior
Myocardial
22 (3.8)
Myocardial
2 (5.9)
lP
infarction, prior
re
infarction, this admission
admission
122 (22.3)
0.53
98 (17.9)
0.65
-p
admission
ro of
Comorbidities, n (%)
60 (11.0)
0.10
20 (3.7)
0.37
99 (17.1)
8 (23.5)
91 (16.7)
0.34
Hypotension
140 (24.1)
7 (20.6)
133 (24.4)
0.84
Respiratory
403 (69.5)
17 (50)
386 (70.7)
0.86
na
Arrhythmia
ur
insufficiency
245 (42.2)
15 (44.1)
230 (42.1)
0.96
Hepatic
110 (19.0)
3 (8.8)
107 (19.6)
0.17
101 (17.4)
5 (14.7)
96 (17.6)
0.82
Diabetes mellitus
197 (34.0)
11 (32.4)
186 (34.1)
1
Baseline evidence
195 (33.6)
9 (26.5)
186 (34.1)
0.46
Jo
Renal insufficiency
insufficiency
Metabolic or electrolyte
abnormality
26
27
of motor, cognitive, or functional deficits Acute stroke
27 (4.7)
1 (2.9)
26 (4.8)
1
Favourable
287 (49.5)
26 (76.5)
261 (47.8)
0.001
Pneumonia
158 (27.2)
2 (5.9)
156 (28.6)
0.002
Bacteraemia
38 (6.6)
0 (0)
38 (7.0)
0.16
Metastatic cancer
112 (19.3)
2 (5.9)
2.8 (2.2)
1.7 (1.7)
neurological status 24 h before cardiac
or any blood borne
Charlson comorbidity index (SD)
0.04
2.8 (2.2)
0.002
SD, standard deviation
Jo
ur
na
lP
re
a
110 (20.1)
-p
malignancy
ro of
arrest
27
28
Table 2. Features, interventions, and outcomes of cardiac arrest events All patients Patients Patients Variables (n = 580) with without favourable favourable neurological neurological outcome at outcome at hospital hospital discharge discharge (n = 34) (n = 546) 208 (35.9) 9 (26.5) 199 (36.4) Arrest at night, n (%) 163 (28.1)
Arrest on weekend, n (%)
7 (20.6)
0.43
241 (41.6)
15 (44.1)
General ward
292 (50.3)
12 (35.3)
Others
47 (8.1)
7 (20.6)
Witnessed arrest, n (%)
402 (69.3)
24 (70.6)
Monitored status, n (%)
352 (60.7)
Shockable rhythm, n (%)
77 (13.3)
ro of
Intensive care unit
226 (41.4) 280 (51.3) 40 (7.3)
1
24 (70.6)
328 (60.1)
0.28
11 (32.4)
66 (12.1)
0.002
154 (26.6)
5 (14.7)
149 (27.3)
0.16
70 (12.1)
3 (8.8)
67 (12.3)
0.79
re
-p
378 (69.2)
Critical care interventions in
na
lP
place at time of arrest, n (%)
Antiarrhythmics
0.27
0.02
Arrest location, n (%)
Mechanical ventilation
156 (28.6)
pvalue
249 (42.9)
9 (26.5)
240 (44.0)
0.05
Dialysis
46 (7.9)
1 (2.9)
45 (8.2)
0.51
Pulmonary artery catheter
5 (0.9)
2 (5.9)
3 (0.5)
0.03
Intra-aortic balloon
5 (0.9)
0 (0)
5 (0.9)
1
CPRa duration, min (SDb)
40.6 (43.6)
12.9 (9.9)
42.3 (44.3)
<0.001
Intra-arrest blood glucose
191.5 (122.4)
207.8 (96.2) 190.5 (123.8)
0.18
0 (0)
57 (10.4)
0.04
4 (11.8)
161 (29.5)
0.03
ur
Vasopressors
Jo
pumping
level, mg/dl (SD)
Intra-arrest hypoglycaemia, n 57 (9.8) (%) Dextrose administered
165 (28.4)
28
29
during CPR, n (%) Dextrose amount
8.7 (17.8)
2.1 (5.9)
9.1 (18.2)
0.02
124 (21.4)
2 (5.9)
122 (22.3)
0.03
60 (10.3)
5 (14.7)
55 (10.1)
0.38
7 (1.2)
3 (8.8)
administered during CPR, g (SD) Combined use of dextrose and insulin, n (%) Post-ROSCc interventions, n (%) Extracorporeal membrane
Targeted temperature management
18 (3.3)
<0.001
Sustained ROSC, n (%)
310 (53.4)
34 (100)
276 (50.5)
<0.001
Survival to hospital
64 (11.0)
34 (100)
30 (5.5)
<0.001
SD, standard deviation.
-p
lP
CPR, cardiopulmonary resuscitation.
re
discharge, n (%)
ROSC, return of spontaneous circulation.
Jo
ur
na
c
0.005
7 (20.6)
intervention
b
4 (0.7)
25 (4.3)
Percutaneous coronary
a
ro of
oxygenation
29
30
Table 3. Multiple logistic regression model with favourable neurological outcome at hospital discharge as the dependent variable Independent variablea
Odds
95%
ratio
confidence
p value
interval Primary model CPRb duration
0.90
0.86-0.94
<0.001
Charlson comorbidity index
0.68
0.54-0.85
<0.001
Post-ROSCc percutaneous coronary
6.00
1.71-21.09
0.005
Favourable neurological status 24 h before
3.65
cardiac arrest 3.59
Pneumonia
0.14
Intra-arrest blood glucose level≤150 mg/dl Primary model with interaction term
0.28
1.43-9.31
0.007
1.37-9.38
0.009
0.03-0.68
0.01
-p
Age between 50 and 77 years
ro of
intervention
0.01
0.90
0.87-0.94
<0.001
0.68
0.54-0.86
0.001
6.69
1.84-24.35
0.004
3.20
1.26-8.12
0.01
Age between 50 and 77 years
3.47
1.31-9.20
0.01
Pneumonia
0.12
0.03-0.61
0.01
0.30
0.11-0.80
0.02
CPR duration Charlson comorbidity index
lP
Post-ROSC percutaneous coronary
re
0.11-0.73
intervention
ur
cardiac arrest
na
Favourable neurological status 24 h before
Non-DMd×Intra-arrest blood glucose
Jo
level≤168 mg/dl
Primary model: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.47, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.94, and the Hosmer and Lemeshow goodness-of-fit Chi-Squared test p = 0.97; Primary model with interaction terms: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.47, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.94, and the Hosmer and Lemeshow goodness-of-fit Chi-Squared test p = 1.00 30
31
The display of independent variables is arranged by the order of p value.
b
CPR, cardiopulmonary resuscitation
c
ROSC, return of spontaneous circulation.
d
DM, diabetes mellitus
Jo
ur
na
lP
re
-p
ro of
a
31
32
Table 4. Multiple logistic regression model with survival to hospital discharge as the dependent variable Independent variablea
Odds
95%
ratio
confidence
p value
interval Secondary model 0.91
0.88-0.94
<0.001
Charlson comorbidity index
0.80
0.69-0.93
0.004
Intra-arrest blood glucose level≤162 mg/dl
0.39
0.20-0.76
0.006
Post-ROSCc percutaneous coronary
4.00
1.38-11.62
0.01
intervention Favourable neurological status 24 h before
2.17
cardiac arrest 0.25
Intra-arrest blood glucose level >381 mg/dl Secondary model with interaction term
0.21
1.16-4.06
0.02
0.08-0.78
0.02 0.04
0.91
0.89-0.94
<0.001
0.28
0.13-0.59
<0.001
Charlson comorbidity index
0.79
0.68-0.92
0.002
Post-ROSC percutaneous coronary
3.94
1.31-11.83
0.01
Hepatic insufficiency
na
0.25
0.08-0.80
0.02
DM×Intra-arrest blood glucose level >273
0.15
0.03-0.74
0.02
0.46
0.24-0.91
0.02
Respiratory insufficiency
0.49
0.26-0.94
0.03
CPR duration level≤155 mg/dl
intervention
lP
Non-DMd×Intra-arrest blood glucose
re
0.05-0.91
ur
-p
Hepatic insufficiency
ro of
CPRb duration
mg/dl
Jo
Vasopressors in place at time of arrest
Secondary model: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.41, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.89, and the Hosmer and Lemeshow goodness-of-fit Chi-Squared test p = 1.00; Secondary model with interaction term: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.43, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.90, and the Hosmer and Lemeshow goodness-of-fit 32
33
Chi-Squared test p = 0.76 The display of independent variables is arranged by the order of p value.
b
CPR, cardiopulmonary resuscitation
c
ROSC, return of spontaneous circulation.
d
DM, diabetes mellitus
Jo
ur
na
lP
re
-p
ro of
a
33
PII:
S0300-9572(19)30702-6
DOI:
https://doi.org/10.1016/j.resuscitation.2019.11.012
Reference:
RESUS 8300
To appear in:
Resuscitation
Received Date:
2 August 2019
Revised Date:
14 October 2019
Accepted Date:
16 November 2019
Please cite this article as: Wang C-Hung, Chang W-Tien, Huang C-Hua, Tsai M-Shan, Chou E, Yu P-Hsun, Wu Y-Wen, Chen W-Jone, Associations between intra-arrest blood glucose level and outcomes of adult in-hospital cardiac arrest: A 10-year retrospective cohort study, Resuscitation (2019), doi: https://doi.org/10.1016/j.resuscitation.2019.11.012
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.
1
Associations between intra-arrest blood glucose level and outcomes of adult inhospital cardiac arrest: A 10-year retrospective cohort study
Chih-Hung Wang, MD, PhD1,2; Wei-Tien Chang, MD, PhD1,2; Chien-Hua Huang, MD, PhD1,2; Min-Shan Tsai, MD, PhD1,2; Eric Chou, MD3; Ping-Hsun Yu, MD4; Yen-Wen
Department of Emergency Medicine, National Taiwan University Hospital, Taipei,
-p
1
ro of
Wu, MD, PhD5-7; Wen-Jone Chen, MD, PhD1,2,8
Taiwan
Department of Emergency Medicine, College of Medicine, National Taiwan
re
2
Department of Emergency Medicine, Baylor Scott&White All Saints Medical Center,
na
3
lP
University, Taipei, Taiwan
Fort Worth, Texas, USA
Department of Emergency Medicine, Taipei Hospital, Ministry of Health and Welfare,
ur
4
Jo
New Taipei City, Taiwan 5
Departments of Internal Medicine and Nuclear Medicine, National Taiwan University
Hospital and National Taiwan University College of Medicine, Taipei, Taiwan 6
Department of Nuclear Medicine and Cardiology Division of Cardiovascular Medical
Center, Far Eastern Memorial Hospital, New Taipei City, Taiwan 1
2
7
National Yang-Ming University School of Medicine, Taipei, Taiwan
8
Division of Cardiology, Department of Internal Medicine, National Taiwan University
Hospital and National Taiwan University College of Medicine, Taipei, Taiwan
Address for correspondence and reprints:
ro of
Wen-Jone Chen, MD, PhD No.7, Zhongshan S. Rd., Zhongzheng Dist., Taipei City 100, Taiwan (R.O.C.)
-p
Tel.:+886-2-23123456 ext. 65792
E-mail: [email protected]
re
Fax: +886-2-2322-3150
ur
Abstract
na
the results for publication
lP
The corresponding author had full access to all data and final responsibility to submit
Jo
Aim: We attempted to examine the association between intra-arrest blood glucose (BG) level and outcomes of in-hospital cardiac arrest (IHCA). The interaction between diabetes mellitus (DM) and BG level as well as between dextrose administration and BG level were investigated.
2
3
Methods: This single-centred retrospective study reviewed IHCA patients between 2006 and 2015. Patients with measured intra-arrest BG levels were included. Multivariable logistic regression analyses were conducted. Generalised additive models were used to identify appropriate cut-off points for continuous variables. Interactions
ro of
between independent variables were assessed during the model-fitting process.
Results: Among the 580 included patients, 34 (5.9%) achieved neurologically intact
-p
survival. There were 197 DM patients (34.0%). The mean intra-arrest BG level was
191.5 mg/dl, with 57 patients (9.8%) experiencing hypoglycaemia (BG level≤70 mg/dl).
re
A total of 165 patients (28.4%) received a dextrose injection. An intra-arrest BG
lP
level≤150 mg/dl was inversely associated with favourable neurological outcomes at hospital discharge (odds ratio [OR]: 0.28, 95% confidence interval [CI]: 0.11-0.73; p-
na
value = 0.01). In analyses of interactions, non-DM×BG level≤168 mg/dl was inversely
ur
associated with favourable neurological outcomes (OR: 0.30, 95% CI: 0.11-0.80; p-
Jo
value = 0.02). There were no significant interactions between BG level and dextrose administration.
Conclusion: IHCA patients with intra-arrest BG level≤150 mg/dl had worse neurological recovery. Intra-arrest hypoglycaemia might be a marker of critical illness. 3
4
Dextrose administration was not shown to improve outcomes of IHCA patients with intra-arrest BG level≤150 mg/dl, indicating the need to develop new therapeutics other than dextrose administration for these patients.
Keywords: Glucose; Hypoglycaemia; Diabetes mellitus; In-Hospital cardiac arrest;
lP
re
-p
ro of
Survival; Neurological outcome.
na
Introduction
ur
In the United States, approximately 209,000 patients experience in-hospital cardiac arrest (IHCA) each year [1]. Despite continuing efforts to improve the “chain
Jo
of survival”, IHCA outcomes remain poor. Approximately 24% of IHCA patients survive to hospital discharge; among these patients, about 14% experience significant neurological disability [1]. The 2005 American Heart Association Guidelines for Cardiopulmonary
4
5
Resuscitation and Emergency Cardiovascular Care [2] suggested the providers should identify a factor that may have caused the arrest during cardiopulmonary resuscitation (CPR) and treat it accordingly. Hypoglycaemia had been listed as one of the reversible causes of cardiac arrest in 2005 guidelines [2], but was removed in the 2010 [3] and 2015 [4] guidelines. Whether blood glucose (BG) levels should be measured during
ro of
CPR and whether intra-arrest hypoglycaemia should be treated were not reviewed in the latest 2015 guidelines [4].
-p
Using Get With The Guidelines-Resuscitation (GWTG-R) database, Moskowitz et al. [5] demonstrated that the proportion of IHCA patients receiving dextrose during
re
CPR has increased from 2001 to 2016 despite that the guidelines [3, 4] had removed
lP
the recommendations regarding dextrose administration. As compared to 2001, patients
na
in 2016 had nearly three-fold higher odds of receiving dextrose during CPR [5]. Also using GWTG-R database, Peng et al.[6] indicated that IHCA patients receiving
ur
dextrose administration during CPR had worse neurological and survival outcomes.
Jo
In the studies by Moskowitz et al.[5] and by Peng et al.[6], the actual BG level during CPR was not measured and consequently its confounding effect was not adjusted in the analysis. Hypoglycaemia had been shown to be associated with increased mortality in critically ill patients [7]. Without accounting for BG level in the analysis, the association between dextrose administration and IHCA outcomes may be 5
6
overestimated. Since dextrose has been used increasingly during CPR [5], there is an urgent need to verify its potential harmful effects. Previous studies [8, 9] indicated that optimal BG levels for post-cardiac arrest patients may differ according to the presence of diabetes mellitus (DM). Therefore, in current studies, we attempted not only to elucidate the association between intra-arrest
ro of
BG level and IHCA outcomes, but also the interactions between DM and BG level or
-p
between dextrose administration and BG level.
Materials and Methods
re
Setting
lP
This retrospective cohort study was performed in the tertiary medical centre,
na
National Taiwan University Hospital (NTUH). NTUH has 2,600 beds, including 220 beds in intensive care units (ICUs). This study was conducted in accordance with the
ur
Declaration of Helsinki amendments. The Research Ethics Committee of the NTUH
Jo
approved this study (reference number: 201805098RINC) and waived the requirement for informed consent before data collection. According to hospital policy, a resuscitation team is activated when a cardiac arrest event occurs in the general wards. A resuscitation team consists of a senior resident, several junior residents, a respiratory therapist, a head nurse, and several ICU nurses. Each code team member is certified to 6
7
provide advanced cardiac life support and capable of offering CPR according to current resuscitation guidelines. For cardiac arrest events in the ICUs, a code team is not mobilised since a sufficient number of experienced staffs is always present in the ICUs. In this case, resuscitation is performed by the staff of the ICU where the cardiac arrest
ro of
event occurred and by staff from neighbouring ICUs.
Participants
-p
Patients who experienced IHCA at NTUH from 2006 to 2015 were screened. Patients who met the following criteria were included in the study: (1) aged 18 years or
re
older, (2) documented absence of pulse with performance of chest compressions for at
lP
least 2 minutes, (3) no documentation of a do-not-resuscitate order before arrest, and (4) measurement of BG level during CPR. Sustained return of spontaneous circulation
na
(ROSC) was defined as ROSC lasting ≥ 20 minutes without resumption of chest
ur
compressions. If multiple cardiac arrest events occurred in a single patient during
Jo
hospitalization, only the first event was recorded. Patients who experienced cardiac arrest related to major trauma were excluded from the study.
Data Collection and Outcome Measures The following information was recorded for each patient: age, gender, 7
8
comorbidities [10], variables derived from the Utstein template [11], the first intraarrest BG level measured during CPR, the amount of dextrose administered during CPR, and critical interventions implemented at the time of cardiac arrest and after sustained ROSC. Duration of CPR was defined as the time interval from the first chest compression initiated by the resuscitation team or ICU members to the termination of
ro of
resuscitation efforts, either due to sustained ROSC or declaration of death. Intra-arrest BG level was measured by blood gas analysers. Hypoglycaemia was defined as a BG
-p
level lower than 70 mg/dL (3.9 mmol/L) [12].
The primary outcome was favourable neurological status at hospital discharge.
re
Favourable neurological status was defined as a score of 1 or 2 on the Cerebral
lP
Performance Category (CPC) scale [13]. Patients with a CPC score of 1 or 2 had
na
sufficient cerebral function to live independently. The CPC score was retrospectively determined by reviewing medical records for each patient. The secondary outcome was
Jo
ur
survival to hospital discharge.
Statistical Analysis R 3.3.1 software (R Foundation for Statistical Computing, Vienna, Austria) was
used for data analysis. Categorical data were expressed as counts and proportions; continuous data were expressed as means and standard deviations. Categorical 8
9
variables were compared using the Fisher’s exact test, and continuous variables were examined by the Wilcoxon rank-sum test. A two-tailed p-value of < 0.05 was considered statistically significant. The odds ratio (OR) was selected as the outcome measure, and multivariable logistic regression analyses were performed to examine the associations between the
ro of
independent variables and outcomes. All available independent variables were considered in the regression model, regardless of whether they were significant by
-p
univariate analysis. The stepwise variable selection procedure (with iterations between
the forward and backward steps) was applied to obtain the final regression model.
re
Significance levels for entry and stay were set at 0.15 to avoid exclusion of potential
lP
candidate variables. The final regression model was identified by sequentially
na
excluding individual variables with a p-value > 0.05 until all regression coefficients were statistically significant.
ur
We used generalised additive models (GAMs) [14] to examine the nonlinear
Jo
effects of continuous variables and, if necessary, to identify the appropriate cut-off point(s) for dichotomising a continuous variable during the variable selection procedure. The interaction between DM and BG level or between dextrose administration and BG level was assessed during the model-fitting process. We assessed the goodness-of-fit of the fitted regression model using c statistics, adjusted generalised R2, and the Hosmer9
10
Lemeshow goodness-of-fit test.
Results A total of 1,698 adult non-trauma patients at NTUH received CPR for ≥ 2 min between 2006 and 2015. Of these, 1,118 patients were excluded because of a lack of
ro of
measurement of BG level during CPR. The remaining 580 patients were included for analysis.
-p
Tables 1 and 2 provide the features of cardiac arrest events before, during, and after CPR for all patients in the cohort. The mean age of the patients was 65.6 years.
re
There were 197 diabetic patients (34.0%). A total of 241 cardiac arrest events (41.6%)
lP
occurred in the ICUs, and 292 events (50.3%) occurred on the general wards. The
na
average CPR duration was 40.6 min. The mean intra-arrest BG level was 191.5 mg/dl, with 57 patients (9.8%) having hypoglycaemia. A total of 165 patients (28.4%) received
ur
injections of 50% dextrose in water, with an average of 8.7 g dextrose being
Jo
administered. Only 64 patients (11.0%) survived to hospital discharge; of these, 34 patients (5.9%) demonstrated favourable neurological status. No patients with hypoglycaemia achieved neurologically intact survival in our cohort. We placed all independent variables listed in Tables 1 and 2 in the regression analysis for variable selection. The GAM plots demonstrated a non-linear association 10
11
of logit (p), where p represented the probability for favourable neurological status (Figure 1) or survival (Supplemental Figure 1), with BG level. If logit (p) was greater than zero, the odds for either outcome was greater than one. Therefore, BG level was transformed into a binary variable during the model-fitting process according to the identified cut-off points.
ro of
As shown in Tables 3 and 4, a BG level≤150 mg/dl was inversely associated with favourable neurological outcome (primary model, OR: 0.28, 95% confidence interval
-p
[CI]: 0.11-0.73; p-value = 0.01) and a BG level≤162 mg/dl was inversely associated
with survival (secondary model, OR: 0.39, 95% CI: 0.20-0.76; p-value = 0.006). The
re
analyses of interactions indicated that non-DM×BG level≤168 mg/dl was inversely
lP
associated with the favourable neurological outcome (primary model with interaction
na
term, OR: 0.30, 95% CI: 0.11-0.80; p-value = 0.02; Table 3) and non-DM×BG level≤155 mg/dl was inversely associated with survival (secondary model with
ur
interaction term, OR: 0.28, 95% CI: 0.13-0.59; p-value <0.001; Table 4). There were
Jo
no significant interactions between dextrose administration and BG levels.
Discussion
Main Findings In univariate analysis, we noted that intra-arrest hypoglycaemia (BG level≤70 11
12
mg/dL) was a significant prognosticator of poor neurological outcome. No patients with hypoglycaemia achieved neurologically intact survival in our cohort. In multivariate analysis, a relatively low intra-arrest BG level (BG level≤150 mg/dl), identified by GAM plot (Figure 1), was found to be associated with worse neurological outcomes. When the confounding effects of BG levels were adjusted, dextrose administration was
ro of
not associated with IHCA outcomes, even for patients with relatively low BG levels. In contrast, for non-DM patients, relatively low BG levels were more significantly
re
Comparison with Previous Studies
-p
associated with worse outcomes, in comparison with DM patients.
lP
The practice of empirical dextrose administration in patients with suspected
na
hypoglycaemia-related events, such as altered mental status or cardiac arrest, was based not only on the concerns that delayed treatment may lead to irreversible brain damage,
ur
but also on the unfounded assumption that the dextrose administration was harmless to
Jo
patients who were normoglycemic or hyperglycaemic [15]. Empirical administration of 25 g dextrose (50 mL of 50% dextrose) without waiting for laboratory confirmation of hypoglycaemia has long been recommended for any patient presenting with coma of undetermined aetiology [15]. Nonetheless, there was substantial evidence questioning this practice [16]. Dextrose administration prior to complete cerebral ischemia or during 12
13
severe incomplete ischemia in experimental animals worsened histologic and neurological outcomes [16]. In a randomised-controlled trial, Longstreth et al. [17] randomised 748 patients with out-of-hospital cardiac arrest into two groups according to the intravenous maintenance fluids, either 5% dextrose in water or normal saline. The results indicated neurological outcomes of patients with out-of-hospital cardiac
ro of
arrest did not differ between the two groups, which suggested that dextrose may still be administered to patients during CPR [17]. Nonetheless, the amount of dextrose given
of administering 25 g dextrose during CPR.
-p
in the trial by Longstreth et al. [17] may be too small to be generalized to the practice
re
By using the large GWTG-R registry data, Peng et al.[6] demonstrated that
lP
dextrose was administered to 4173 (4.2%) IHCA patients. Both multivariate regression
na
analysis and propensity-matched analysis indicated that dextrose administration was inversely associated with favourable neurological and survival outcomes [6].
ur
Nevertheless, BG level was not accounted for in the study by Peng et al.[6]. Also
Jo
analysing the data from GWTG-R, Beiser et al.[9] indicated that non-DM patients were sensitive to hypoglycaemia (BG level<70 mg/dl) during the early post-ROSC periods. For other critically ill patients, hypoglycaemia (BG level<70 mg/dl) has also been associated with mortality in a dose-dependent manner, i.e. the lower the BG level, the higher the mortality was [7]. In our cohort, we also noted that for patients with intra13
14
arrest hypoglycaemia (BG level≤70 mg/dl), none regained favourable neurological status. Since hypoglycaemia was one of the indications for administering dextrose [15], without controlling the effects of hypoglycaemia or BG levels, the effects of dextrose administration may be biased by confounding by indications, leading to overestimated associations.
ro of
As demonstrated in our analysis, when the BG level was accounted for in the multivariate analysis, neither dextrose administration nor the amount of dextrose given
-p
was associated with IHCA outcomes. Categorizing a continuous variable without knowing the relationship between independent and dependent variables may lead to
re
loss of information and statistical power [18]. Therefore, we used GAM plots (Figure
lP
1 & Supplemental Figure 1) to identify the optimal cut-off points for BG levels and put
na
the transformed BG levels into multivariate regression analyses for variable selection along with the variable of intra-arrest hypoglycaemia. The results showed intra-arrest
ur
BG levels were significantly associated with outcomes, especially when the BG level
Jo
was relatively low. Interestingly, the upper limit of the relatively low BG level, 150 mg/dl, was over twice the threshold defining hypoglycaemia (70 mg/dl).
Prognostic Role of Intra-arrest BG level In the 2005 resuscitation guidelines [2], hypoglycaemia was listed as one of the 14
15
treatable or reversible causes of cardiac arrest and thus, hypoglycaemia should be identified and treated promptly. Nonetheless, how hypoglycaemia leads to cardiac arrest and whether dextrose administration could reverse this pathophysiological process had not been clearly elucidated in previous studies. In a rat study, Reno et al. [19] demonstrated that severe hypoglycaemia (10-15 mg/dl) induced various cardiac
ro of
arrhythmias, such as premature ventricular contractions, tachycardia, and high-degree heart block. When beta-adrenergic blockers were administered, cardiac arrhythmias
-p
were reduced and hypoglycaemia-related deaths were completely prevented, which
suggested that an overactive sympathoadrenal response may partly account for death
re
related to hypoglycaemia [19]. Therefore, it seemed that dextrose administration alone
na
analyses of interactions.
lP
may not reverse this hypothesized pathophysiological process, as demonstrated in our
Alternatively, hypoglycaemia may just be an epiphenomenon of critical illness. In
ur
a post hoc analysis of the NICE-SUGAR trial [7], investigators revealed that the
Jo
association between hypoglycaemia and mortality was attenuated after adjusting the confounding effects of baseline characteristics and post-randomization factors. Interestingly, the investigators of the NICE-SUGAR trial [7] also reported that for patients not being treated with insulin, the association between hypoglycaemia and mortality was greater, and the time to death was shorter, compared with those receiving 15
16
insulin. The presence of so-called spontaneous hypoglycaemia, i.e. hypoglycaemia not caused by insulin administration, and its strong association with poor outcome highly implied that hypoglycaemia may just be a disease marker and consequently, the hypoglycaemia-targeted therapeutics may not be successful [7]. In our analyses of interactions, we also noted that non-DM patients were more sensitive to relatively low
ro of
BG levels compared with DM patients. Therefore, hypoglycaemia, or relatively low BG levels, may not be a treatable cause; instead, these test results could be viewed as
-p
prognosticators, indicating illness severity. Treatment should thus be focused on
searching for other potential causes leading to cardiac arrest, rather than supplementing
lP
na
Future Applications
re
with dextrose only.
Transient hyperglycaemia during critical illness in patients without DM had been
ur
assumed to be harmless or even advantageous [20]. The development of stress
Jo
hyperglycaemia was thought to be the result of a highly complicated interplay of counter-regulatory hormones, such as catecholamines and cortisol [21, 22]. The absence of stress hyperglycaemia was thought to be caused by a failure in activation of the hypothalamic-pituitary-adrenal axis [20]. As revealed by the rat study of Reno et al. [17], plasma epinephrine levels increased shortly after the onset of severe 16
17
hypoglycaemia with early peaking and then decreasing, possibly indicating adrenal exhaustion. Several studies [23] have also shown that relative adrenal insufficiency was common during the post-ROSC period. Therefore, relatively low intra-arrest BG levels may be an early manifestation of unrecognized relative adrenal insufficiency. For suspected adrenal insufficiency, glucocorticoid administration may be
ro of
considered to improve the haemodynamic response. In a randomized controlled trial by Mentzelopoulos et al.[24] for IHCA, patients allocated to the experimental group
-p
received vasopressin and methylprednisolone during CPR and hydrocortisone for postROSC shock; these patients experienced significantly better outcomes than those in the
re
control group. Nonetheless, in the study by Mentzelopoulos et al.[24], the intra-arrest
lP
BG levels of the two groups did not differ significantly; therefore, whether
na
methylprednisolone supplementation during CPR can improve outcomes for patients with relatively low intra-arrest BG levels should be further examined. For post-ROSC
ur
patients, an evaluation for relative adrenal insufficiency was not frequently performed
Jo
even if they were in shock status [25]. Relatively low intra-arrest BG levels might serve as a sign of relative adrenal insufficiency, reminding clinicians that diagnostic procedures or empirical treatments for relative adrenal insufficiency should be performed.
17
18
Study Limitations First, our study was based on a highly selected cohort, including only 34.2% (580/1698) of all screened patients. Most patients were excluded because of a lack of intra-arrest BG measurement, probably caused by removal of hypoglycaemia from the list of reversible causes of cardiac arrest in resuscitation guidelines [4]. Second, this
ro of
was an observational study, which can only establish an association, rather than a causal relationship, between independent and dependent variables. The exact mechanisms
-p
underlying the association between relatively low intra-arrest BG levels and less favourable outcomes could not be examined in the current analysis. Although relative
re
adrenal insufficiency was suspected to account for this association, the levels of ACTH
lP
and cortisol were not measured in most included patients, as reported in previous
na
studies [25]. Third, the effects of unmeasured confounders may have biased the results despite having used multivariate analysis to adjust the effects of measured confounding
ur
factors. For example, pre-arrest events leading to the relatively low BG levels, pre-
Jo
arrest medication status regarding diabetic control, and the reasons why intra-arrest BG levels were measured may be important confounding factors and could not be verified in a retrospective manner. These limitations can only be solved by a prospective study. Our results may be deemed as hypothesis-generating and serve as the foundation for future study. For example, factors listed in Tables 3 and 4 may be used to design the 18
19
data collection form in the prospective study.
Conclusions Relatively low intra-arrest BG levels (BG≤150 mg/dl) were inversely associated with favourable neurological status at hospital discharge. Dextrose administration was
ro of
not shown to improve outcomes of IHCA patients with relatively low BG levels. For non-DM patients, relatively low BG levels were more significantly associated with
-p
worse outcomes, in comparison with DM patients. These results implied that relatively low intra-arrest BG levels may simply be a disease marker; thus, newer therapeutics
na
lP
re
other than dextrose administration should be developed.
Conflicts of Interests
Jo
ur
The authors declare that they have no conflict of interest.
Acknowledgments We thank Centre of Quality Management of National Taiwan University Hospital for providing the list of patients sustaining in-hospital cardiac arrest. We thank the staff of the 3rd Core Lab, Department of Medical Research, National Taiwan University 19
20
Hospital for technical support. Author Chih-Hung Wang recieved a grant (108-S4091) from the National Taiwan University Hospital. Author Wen-Jone Chen recieved a grant (108-S4236) from the National Taiwan University Hospital. National Taiwan University Hospital had no involvement in designing the study, collecting, analysing or interpreting the data, writing the manuscript, or deciding whether to submit the
Jo
ur
na
lP
re
-p
ro of
manuscript for publication.
20
21
References 1.
Benjamin EJ, Blaha MJ, Chiuve SE, Cushman M, Das SR, Deo R, et al. Heart Disease and Stroke Statistics-2017 Update: A Report From the American Heart Association. Circulation 2017; 135:e146-e603. Part 7.2: Management of Cardiac Arrest. Circulation 2005; 112:IV-58-IV-66.
3.
Neumar RW, Otto CW, Link MS, Kronick SL, Shuster M, Callaway CW, et al.
ro of
2.
Part 8: adult advanced cardiovascular life support: 2010 American Heart
-p
Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2010; 122:S729-67.
Link MS, Berkow LC, Kudenchuk PJ, Halperin HR, Hess EP, Moitra VK, et al.
re
4.
lP
Part 7: Adult Advanced Cardiovascular Life Support: 2015 American Heart
na
Association Guidelines Update for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2015; 132:S444-64. Moskowitz A, Ross CE, Andersen LW, Grossestreuer AV, Berg KM, Donnino
ur
5.
Jo
MW, et al. Trends Over Time in Drug Administration During Adult In-Hospital Cardiac Arrest. Crit Care Med 2019; 47:194-200.
6.
Peng TJ, Andersen LW, Saindon BZ, Giberson TA, Kim WY, Berg K, et al. The administration of dextrose during in-hospital cardiac arrest is associated with increased mortality and neurologic morbidity. Crit Care 2015; 19:160. 21
22
7.
NICE-SUGAR Study Investigators, Finfer S, Liu B, Chittock DR, Norton R, Myburgh JA, et al. Hypoglycemia and risk of death in critically ill patients. N Engl J Med 2012: 367:1108-18.
8.
Wang CH, Huang CH, Chang WT, Tsai MS, Yu PH, Wu YW, et al. Associations between blood glucose level and outcomes of adult in-hospital
9.
ro of
cardiac arrest: a retrospective cohort study. Cardiovasc Diabetol 2016; 15:118. Beiser DG, Carr GE, Edelson DP, Peberdy MA, Hoek TL. Derangements in
-p
blood glucose following initial resuscitation from in-hospital cardiac arrest: a
Resuscitation 2009; 80:624-30.
Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of
lP
10.
re
report from the national registry of cardiopulmonary resuscitation.
na
classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40: 373-83. Jacobs I, Nadkarni V, Bahr J, Berg RA, Billi JE, Bossaert L, et al. Cardiac arrest
ur
11.
Jo
and cardiopulmonary resuscitation outcome reports: update and simplification of the Utstein templates for resuscitation registries: a statement for healthcare professionals from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian Resuscitation Council, New Zealand Resuscitation Council, Heart 22
23
and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Councils of Southern Africa). Circulation 2004; 110:3385-97. 12.
Workgroup on Hypoglycemia, American Diabetes Association. Defining and reporting hypoglycemia in diabetes: a report from the American Diabetes Association Workgroup on Hypoglycemia. Diabetes Care 2005; 28:1245-9. Becker LB, Aufderheide TP, Geocadin RG, Callaway CW, Lazar RM, Donnino
ro of
13.
MW, et al. Primary outcomes for resuscitation science studies: a consensus
-p
statement from the American Heart Association. Circulation 2011; 124:215877.
Hastie TJ, Tibshirani RJ. Generalized Additive Models. London and New York:
lP
Chapman & Hall; 1990.
re
14.
Posner JB. The comatose patient. JAMA 1975; 233:1313-4.
16.
Browning RG, Olson DW, Stueven HA, Mateer JR. 50% dextrose: antidote or
na
15.
Longstreth WT Jr, Copass MK, Dennis LK, Rauch-Matthews ME, Stark MS,
Jo
17.
ur
toxin? Ann Emerg Med 1990; 19:683-7.
Cobb LA. Intravenous glucose after out-of-hospital cardiopulmonary arrest: a community-based randomized trial. Neurology 1993; 43:2534-41.
18.
Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ 2006; 332:1080. 23
24
19.
Reno CM, Daphna-Iken D, Chen YS, VanderWeele J, Jethi K, Fisher SJ. Severe hypoglycemia-induced
lethal
cardiac
arrhythmias
are
mediated
by
sympathoadrenal activation. Diabetes 2013; 62:3570-81. 20.
Dungan KM, Braithwaite SS, Preiser JC. Stress hyperglycaemia. Lancet 2009; 373:1798-807. Barth E, Albuszies G, Baumgart K, Matejovic M, Wachter U, Vogt J, et al.
ro of
21.
Glucose metabolism and catecholamines. Crit Care Med 2007; 35:S508-S18.
Andrews RC, Walker BR. Glucocorticoids and insulin resistance: old hormones,
-p
22.
new targets. Clin Sci (Lond) 1999; 96:513-23.
Pene F, Hyvernat H, Mallet V, Cariou A, Carli P, Spaulding C, et al. Prognostic
re
23.
lP
value of relative adrenal insufficiency after out-of-hospital cardiac arrest.
24.
na
Intensive Care Med 2005; 31:627-33.
Mentzelopoulos SD, Malachias S, Chamos C, Konstantopoulos D, Ntaidou T,
ur
Papastylianou A, et al. Vasopressin, steroids, and epinephrine and
Jo
neurologically favorable survival after in-hospital cardiac arrest: a randomized clinical trial. JAMA 2013; 310:270-9.
25.
Miller JB, Donnino MW, Rogan M, Goyal N. Relative adrenal insufficiency in post-cardiac arrest shock is under-recognized. Resuscitation 2008; 76:221-5.
24
25
Figure Legends Figure 1. Generalised additive model plot for nonparametric modelling of the effect of intra-arrest blood glucose level on the logit of probability for favourable neurological
Jo
ur
na
lP
re
-p
ro of
outcomes at hospital discharge
25
26
Table 1. Baseline characteristics of study patients Variables All patients Patients with (n = 580) favourable neurological outcome at hospital discharge (n = 34)
Patients without favourable neurological outcome at hospital discharge (n = 546)
pvalue
Age, y (SDa)
65.6 (16.1)
64.0 (13.0)
65.7 (16.3)
0.33
Male, n (%)
330 (56.9)
23 (67.6)
307 (56.2)
0.22
131 (22.6)
9 (26.5)
105 (18.1)
7 (20.6)
67 (11.6)
7 (20.6)
Heart failure, this admission Heart failure, prior
Myocardial
22 (3.8)
Myocardial
2 (5.9)
lP
infarction, prior
re
infarction, this admission
admission
122 (22.3)
0.53
98 (17.9)
0.65
-p
admission
ro of
Comorbidities, n (%)
60 (11.0)
0.10
20 (3.7)
0.37
99 (17.1)
8 (23.5)
91 (16.7)
0.34
Hypotension
140 (24.1)
7 (20.6)
133 (24.4)
0.84
Respiratory
403 (69.5)
17 (50)
386 (70.7)
0.86
na
Arrhythmia
ur
insufficiency
245 (42.2)
15 (44.1)
230 (42.1)
0.96
Hepatic
110 (19.0)
3 (8.8)
107 (19.6)
0.17
101 (17.4)
5 (14.7)
96 (17.6)
0.82
Diabetes mellitus
197 (34.0)
11 (32.4)
186 (34.1)
1
Baseline evidence
195 (33.6)
9 (26.5)
186 (34.1)
0.46
Jo
Renal insufficiency
insufficiency
Metabolic or electrolyte
abnormality
26
27
of motor, cognitive, or functional deficits Acute stroke
27 (4.7)
1 (2.9)
26 (4.8)
1
Favourable
287 (49.5)
26 (76.5)
261 (47.8)
0.001
Pneumonia
158 (27.2)
2 (5.9)
156 (28.6)
0.002
Bacteraemia
38 (6.6)
0 (0)
38 (7.0)
0.16
Metastatic cancer
112 (19.3)
2 (5.9)
2.8 (2.2)
1.7 (1.7)
neurological status 24 h before cardiac
or any blood borne
Charlson comorbidity index (SD)
0.04
2.8 (2.2)
0.002
SD, standard deviation
Jo
ur
na
lP
re
a
110 (20.1)
-p
malignancy
ro of
arrest
27
28
Table 2. Features, interventions, and outcomes of cardiac arrest events All patients Patients Patients Variables (n = 580) with without favourable favourable neurological neurological outcome at outcome at hospital hospital discharge discharge (n = 34) (n = 546) 208 (35.9) 9 (26.5) 199 (36.4) Arrest at night, n (%) 163 (28.1)
Arrest on weekend, n (%)
7 (20.6)
0.43
241 (41.6)
15 (44.1)
General ward
292 (50.3)
12 (35.3)
Others
47 (8.1)
7 (20.6)
Witnessed arrest, n (%)
402 (69.3)
24 (70.6)
Monitored status, n (%)
352 (60.7)
Shockable rhythm, n (%)
77 (13.3)
ro of
Intensive care unit
226 (41.4) 280 (51.3) 40 (7.3)
1
24 (70.6)
328 (60.1)
0.28
11 (32.4)
66 (12.1)
0.002
154 (26.6)
5 (14.7)
149 (27.3)
0.16
70 (12.1)
3 (8.8)
67 (12.3)
0.79
re
-p
378 (69.2)
Critical care interventions in
na
lP
place at time of arrest, n (%)
Antiarrhythmics
0.27
0.02
Arrest location, n (%)
Mechanical ventilation
156 (28.6)
pvalue
249 (42.9)
9 (26.5)
240 (44.0)
0.05
Dialysis
46 (7.9)
1 (2.9)
45 (8.2)
0.51
Pulmonary artery catheter
5 (0.9)
2 (5.9)
3 (0.5)
0.03
Intra-aortic balloon
5 (0.9)
0 (0)
5 (0.9)
1
CPRa duration, min (SDb)
40.6 (43.6)
12.9 (9.9)
42.3 (44.3)
<0.001
Intra-arrest blood glucose
191.5 (122.4)
207.8 (96.2) 190.5 (123.8)
0.18
0 (0)
57 (10.4)
0.04
4 (11.8)
161 (29.5)
0.03
ur
Vasopressors
Jo
pumping
level, mg/dl (SD)
Intra-arrest hypoglycaemia, n 57 (9.8) (%) Dextrose administered
165 (28.4)
28
29
during CPR, n (%) Dextrose amount
8.7 (17.8)
2.1 (5.9)
9.1 (18.2)
0.02
124 (21.4)
2 (5.9)
122 (22.3)
0.03
60 (10.3)
5 (14.7)
55 (10.1)
0.38
7 (1.2)
3 (8.8)
administered during CPR, g (SD) Combined use of dextrose and insulin, n (%) Post-ROSCc interventions, n (%) Extracorporeal membrane
Targeted temperature management
18 (3.3)
<0.001
Sustained ROSC, n (%)
310 (53.4)
34 (100)
276 (50.5)
<0.001
Survival to hospital
64 (11.0)
34 (100)
30 (5.5)
<0.001
SD, standard deviation.
-p
lP
CPR, cardiopulmonary resuscitation.
re
discharge, n (%)
ROSC, return of spontaneous circulation.
Jo
ur
na
c
0.005
7 (20.6)
intervention
b
4 (0.7)
25 (4.3)
Percutaneous coronary
a
ro of
oxygenation
29
30
Table 3. Multiple logistic regression model with favourable neurological outcome at hospital discharge as the dependent variable Independent variablea
Odds
95%
ratio
confidence
p value
interval Primary model CPRb duration
0.90
0.86-0.94
<0.001
Charlson comorbidity index
0.68
0.54-0.85
<0.001
Post-ROSCc percutaneous coronary
6.00
1.71-21.09
0.005
Favourable neurological status 24 h before
3.65
cardiac arrest 3.59
Pneumonia
0.14
Intra-arrest blood glucose level≤150 mg/dl Primary model with interaction term
0.28
1.43-9.31
0.007
1.37-9.38
0.009
0.03-0.68
0.01
-p
Age between 50 and 77 years
ro of
intervention
0.01
0.90
0.87-0.94
<0.001
0.68
0.54-0.86
0.001
6.69
1.84-24.35
0.004
3.20
1.26-8.12
0.01
Age between 50 and 77 years
3.47
1.31-9.20
0.01
Pneumonia
0.12
0.03-0.61
0.01
0.30
0.11-0.80
0.02
CPR duration Charlson comorbidity index
lP
Post-ROSC percutaneous coronary
re
0.11-0.73
intervention
ur
cardiac arrest
na
Favourable neurological status 24 h before
Non-DMd×Intra-arrest blood glucose
Jo
level≤168 mg/dl
Primary model: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.47, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.94, and the Hosmer and Lemeshow goodness-of-fit Chi-Squared test p = 0.97; Primary model with interaction terms: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.47, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.94, and the Hosmer and Lemeshow goodness-of-fit Chi-Squared test p = 1.00 30
31
The display of independent variables is arranged by the order of p value.
b
CPR, cardiopulmonary resuscitation
c
ROSC, return of spontaneous circulation.
d
DM, diabetes mellitus
Jo
ur
na
lP
re
-p
ro of
a
31
32
Table 4. Multiple logistic regression model with survival to hospital discharge as the dependent variable Independent variablea
Odds
95%
ratio
confidence
p value
interval Secondary model 0.91
0.88-0.94
<0.001
Charlson comorbidity index
0.80
0.69-0.93
0.004
Intra-arrest blood glucose level≤162 mg/dl
0.39
0.20-0.76
0.006
Post-ROSCc percutaneous coronary
4.00
1.38-11.62
0.01
intervention Favourable neurological status 24 h before
2.17
cardiac arrest 0.25
Intra-arrest blood glucose level >381 mg/dl Secondary model with interaction term
0.21
1.16-4.06
0.02
0.08-0.78
0.02 0.04
0.91
0.89-0.94
<0.001
0.28
0.13-0.59
<0.001
Charlson comorbidity index
0.79
0.68-0.92
0.002
Post-ROSC percutaneous coronary
3.94
1.31-11.83
0.01
Hepatic insufficiency
na
0.25
0.08-0.80
0.02
DM×Intra-arrest blood glucose level >273
0.15
0.03-0.74
0.02
0.46
0.24-0.91
0.02
Respiratory insufficiency
0.49
0.26-0.94
0.03
CPR duration level≤155 mg/dl
intervention
lP
Non-DMd×Intra-arrest blood glucose
re
0.05-0.91
ur
-p
Hepatic insufficiency
ro of
CPRb duration
mg/dl
Jo
Vasopressors in place at time of arrest
Secondary model: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.41, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.89, and the Hosmer and Lemeshow goodness-of-fit Chi-Squared test p = 1.00; Secondary model with interaction term: Goodness-of-fit assessment: n = 580, adjusted generalized R2 = 0.43, the estimated area under the Receiver Operating Characteristic (ROC) curve = 0.90, and the Hosmer and Lemeshow goodness-of-fit 32
33
Chi-Squared test p = 0.76 The display of independent variables is arranged by the order of p value.
b
CPR, cardiopulmonary resuscitation
c
ROSC, return of spontaneous circulation.
d
DM, diabetes mellitus
Jo
ur
na
lP
re
-p
ro of
a
33