Development a clinical prediction model of the neurological outcome for patients with coma and survived 24 hours after cardiopulmonary resuscitation

Abstract Background Cardiac arrest is still a global public health problem at present. The neurological outcome is the core indicator of the prognosis of cardiac arrest. However, there is no effective means or tools to predict the neurological outcome of patients with coma and survived 24 hours after successful cardiopulmonary resuscitation (CPR). Hypothesis Therefore, we expect to construct a prediction model to predict the neurological outcome for patients with coma and survived 24 hours after successful CPR. Methods A retrospective cohort study was used to construct a prediction model of the neurological function for patients with coma and survived 24 hours after successful CPR. From January 2007 to December 2015, a total of 262 patients met the inclusion and exclusion criteria. Results The predictive model was developed using preselected variables by a systematic review of the literature. Finally, we get five sets of models (three sets of construction models and two sets of internal verification models) which with similar predictive value. The stepwise model, which including seven variables (age, noncardiac etiology, nonshockable rhythm, bystander CPR, total epinephrine dose, APTT, and SOFA score), was the simplest model, so we choose it as our final predictive model. The area under the ROC curve (AUC), specificity, and sensitivity of the stepwise model were respectively 0.82 (0.77, 0.87), 0.72and 0.82. The AUC, specificity, and sensitivity of the bootstrap stepwise (BS stepwise) model were respectively 0.82 (0.77, 0.87), 0.71, and 0.82. Conclusion This new and validated predictive model may provide individualized estimates of neurological function for patients with coma and survived 24 hours after successful CPR using readily obtained clinical risk factors. External validation studies are required further to demonstrate the model's accuracy in diverse patient populations.


| BACKGROUND
Cardiac arrest is still a global public health problem at present. The worldwide incidence rate of cardiac arrest is about 418/100 000. Cardiac arrest causes 15% to 20% of global deaths every year. 1,2 With the development and popularization of cardiopulmonary resuscitation (CPR), the restoration of spontaneous circulation (ROSC) has been dramatically improved, but the overall survival rate is still only about 10%. The neurological outcome is the core indicator of the prognosis of cardiac arrest. 3 Among the patients with successful CPR, the rate of favorable neurological outcome was only 2.7% to 13.9%. [4][5][6] Most of the patients with unfavorable neurological outcome eventually died or were in a vegetative state. 7,8 At present, related studies had shown that electroencephalogram, short-delay somatosensory evoked potential (SSEEP), neuron-specific enolase (NSE), S-100B, microRNAs (miRNAs), near-infrared spectroscopy (NIRS), and so forth may be used as a predictor of neurological function for patients with successful CPR. 9,10 The prediction of outcomes at 24 hours after ROSC may be clinically less important since patients after CPR. However, the 24-hour survival of CPR is only about 10%, and their final prognosis is also uncertain. Only one third or fewer patients who survived 24 hours after successful CPR could discharge. 11 At the same time, there is no effective means or tools to predict the neurological outcome of patients with coma and survived 24 hours after successful CPR. Patients with coma and survived 24 hours after successful CPR. Therefore, we expect to construct a prediction tool that may be used to predict the neurological outcome for patients with coma and survived 24 hours after successful CPR.

| Study design
A retrospective cohort study.

| Objective
To construct a prediction model that could predict the favorable neurological outcome of patients with coma and survived 24 hours after successful CPR.

| Data source
The data in this study were provided by Fabio Silvio Taccone

| Inclusion criteria
(a) The patient was in a coma (Glasgow Coma Scale, GCS < 9); (b) Patients survived more than 24 hours after intensive care unit (ICU) admission.

| Exclusion criteria
Patients with previous neurological impairment, including paralysis, muscle weakness, poor coordination, loss of sensation, seizures, confusion, pain, and altered levels of consciousness.

| Participants
From January 2007 to December 2015, a total of 435 patients were with successfully CPR, 51 of whom died within 24 hours, 76 of whom without coma, and ultimately 308 of whom met the inclusion criteria.
Forty-six patients with previous neurological impairment were excluded. Finally, 262 patients met the inclusion and exclusion criteria.

| Postresuscitation care
The protocol of postresuscitation management has been extensively described elsewhere, widely accepted, and applied. Briefly, all cardiac arrest patients with coma received targeted temperature management (TTM; target body temperature: 32-34 C) for 24 hours. Rewarming (<0.5 C/h) was achieved passively. Midazolam, morphine, and cisatracurium were administered for deep sedation to control shivering. PiCCO technology was used to monitor hemodynamics. Repeated transesophageal and transthoracic echocardiography were used to assess cardiac function. Mean arterial pressure (MAP) was maintained at >65 to 70 mm Hg using volume resuscitation, dobutamine, or noradrenaline, whenever needed. Extracorporeal membrane oxygenation (ECMO) or intra-aortic balloon counterpulsation (IABP) was also used in patients with severe cardiogenic shock. Mechanical ventilation was used to maintain SpO2 > 94% and normocapnia. Blood glucose was maintained at 110 to 150 mg/dL by a continuous insulin infusion.
Enteral nutrition was initiated during TTM and continued after that according to gastric tolerance.

| Clinical and biochemical data collection
The following data were collected on the day of admission:

| Selection of predictor variables
A review of literature was performed for the risk factors related to neurological outcomes in patients after CPR. The risk factors considered in the model are those that are deemed to be significantly related to neurological outcomes in patients after CPR and are readily available in clinical practice. Other risk factors may be worth including, but because they can only be measured by time-consuming, expensive, or invasive testing procedures, these risk factors are usually not taken into account.
To make these models as easy to use in clinical practice and minimize noise, we restrict our attention to those risk factors that are generally accepted, readily available, and precisely measured in clinical practice.   The clinical characteristics of patients   Table 3 and Figure 1).

| Verification of prediction model
As the relatively same sample size of our study, we adopted boo-  Age is not only a significant risk factor for cardiac arrest but also an important prognostic factor for patients with cardiac arrest. As one grows older, the risk suffering from diabetes, 14 hypertension, 15 cardiovascular, and cerebrovascular diseases increased. 16 These diseases are also significant risks of cardiac arrest. 17 Cardiac etiology was the most common reason for cardiac arrest, about 50% to 60% of which is induced by heart-related diseases, such as myocardial infarction, arrhythmia, or cardiac failure, and so forth. 18 A retrospective study included 1041 patients with cardiac arrest, 63% of which were with cardiac etiology, and only 37% were with noncardiac etiology. This study found that the prognosis of patients with cardiac etiology was significantly better than the patients with noncardiac etiology (44% vs 23%, P < .01). 19 The possible reasons were: (a) Patients with cardiac etiology were often complicated with heart-related diseases in the past. A more accurate diagnosis could be obtained through a brief medical history, which could be provided by patients themselves, their family members, or their private doctors.
Then the targeted treatment could be given quickly and accurately. could not be quickly relieved, and more importantly, a simple respirator was the first and most frequently received respiratory support devices, which with poor respiratory support. 20 Although relevant studies had shown that initial shockable rhythm was closely associated with the prognosis of patients with cardiac arrest, [21][22][23] We used a nonshockable initial rhythm instead of initial shockable rhythm in our study. The reasons were as follows: (a) nonshockable initial rhythm could be quickly recorded before cardiac arrest occurred because nonshockable initial rhythm taken a relatively long time to develop into cardiac arrest 24 ; (b) the most of the initial shockable rhythm of cardiac arrest were developed from nonshockable initial rhythm 24 ; (c) The time that initial shockable rhythm developed into cardiac arrest was concise, as a result, that the discover and diagnose the initial shockable rhythm with very difficult neither in or out of the hospital. Finally, the initial shockable rhythm could not be recorded adequately; (d) Related studies also showed that nonshockable initial rhythm was an independent risk factor for the prognosis of patients with cardiac arrest. 24 Epinephrine was an essential drug for the treatment of cardiac arrest, and its dosage was strictly related to the prognosis of patients with cardiac arrest. 28,29 In the process of CPR, the more epinephrine, the less response to epinephrine and the more side effects of epinephrine. 30,31 Due to blood stop flowing and endothelial cell hypoxia damage after cardiac arrest, coagulation pathways would be activated, including endogenous and exogenous coagulation pathways, resulting in thrombosis. Thrombosis would further aggravate cerebral ischemia and hypoxia. Therefore, the coagulation status in the body after cardiac arrest was closely related to the neurological function of patients after CPR.
PT or INR was an indicator of the exogenous coagulation pathway, while APTT was an indicator of the endogenous coagulation pathway. 32 Our study found that APTT was firmly related to the neurological prognosis of cardiac arrest. The reason was that the function of the endogenous coagulation pathway was mainly reflected inactivating the coagulation system, while the waterfall reaction of the coagulation system was manifested primarily on the endogenous coagulation pathway.
Besides, platelets were also an essential part of thrombosis.
SOFA score was currently used for organ failure score and was widely used for the prognosis prediction of critically ill patients. [33][34][35] A study including 173 out-of-hospital CPR patients treated with hypothermia found that the higher the SOFA score, the higher the mortality rate and the worse neurological function during hospitalization. 36 MFP can be used to investigate whether the relationship between covariables and independent variables is nonlinear. 37

CONFLICT OF INTEREST
The authors declare no potential conflict of interests.

ETHICS STATEMENT
New ethics approval was not applicable, because the original author had obtained the ethical approval when conducting this study. Permission to participate was also not appropriate, because our review was a retrospective study of data reuse, and the message of the patients was anonymous.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available in [Dryad] at https://datadryad.org/resource/doi:10.5061/dryad.qv6fp83, Reference 13.