In the United States, 420,000 people suffer from out-of-hospital cardiac arrest each year, and the survival rate is only about 6% (1, 2). Early assessment in patients with cardiac arrest can improve their prognosis and reduce the mortality rate (3). Cardiopulmonary resuscitation (CPR) is the only effective treatment for cardiac arrest (4–6). In recent years, remarkable progress has been made in the field of CPR, such as the optimization of compression techniques, the popular application of automated external defibrillators (AEDs), and comprehensive treatment strategies after CPR. Every innovation and improvement may have a significant impact on the survival rate of patients. The widespread dissemination of education and training has enabled more members of the public to master this life-saving skill, thereby providing timely assistance in emergencies. Despite recent progress in resuscitation medicine and critical care medicine, the mortality rate after cardiac arrest remains high, especially in the case of out-of-hospital cardiac arrest (7). Accurate prognosis assessment is an important method for improving treatment efficiency, enhancing outcomes, preserving patient dignity, and reducing the burden of cardiac arrest.

Several studies have developed predictive models for hospital mortality following CPR, yet these models possess certain limitations, including their application primarily to pediatric patients or those experiencing out-of-hospital cardiac arrest, suboptimal prediction accuracy, and the omission of clinical laboratory variables (8–11). Therefore, we aimed to investigate the variables associated with hospital mortality for patients who survived 24 h after CPR, and then construct a predictive model that can guide clinical treatment decisions and improve resuscitation survival.

We included 374 patients who underwent CPR after cardiac arrest from the database, with 262 patients in the training group and 112 patients in the validation group. Our study indicated that age (OR 1.05, 95% CI: 1.03–1.08), witnessed arrest (OR 0.28, 95% CI: 0.11–0.73), time to ROSC (OR 1.05, 95% CI: 1.02–1.08), non-shockable rhythm (OR 3.41, 95% CI: 1.61–7.18), alkaline phosphatase (OR 1.01, 95% CI: 1–1.01) and SOFA (OR 1.27, 95% CI: 1.15–1.4) were independent critical factors for predicting the risk of death 24 h after CPR. Furthermore, we constructed a nomogram based on these selected factors, which demonstrated good discrimination, accuracy, and clinical utility.

In the early stages of cardiac arrest, multiple organs in the body were still in a better state of oxygenation, and shockable rhythms were more likely to be successfully resuscitated at this stage, thereby increasing the proportion of good prognosis. In contrast, non-shockable rhythms often indicated cardiac energy depletion, failure of function, and a low percentage of successful resuscitation (14, 15). Shockable rhythms, including ventricular fibrillation (VF) and pulseless ventricular tachycardia (PVT), were traditionally regarded as arrhythmias that could be reversed. By administering defibrillation, these rhythms can be corrected to restore the heart's normal rhythm and ensure proper blood flow to organs, thus enhancing the likelihood of survival (14, 16). One study showed that despite the presence of refractory out-of-hospital cardiac arrest, patients with shockable rhythms were more likely to have a lower risk of poor prognosis (180-day mortality), with a hazard ratio (HR) of 0.27 and a 95% confidence interval (CI) of 0.18–0.41 (17). Another study showed that patients with shockable rhythms after sudden cardiac arrest had a higher rate of survival discharged from the hospital (OR 5.21, 95% CI: 2.99–9.07) than those with non-shockable rhythms, after adjusting for sex, age, comorbidities, and ethnicity (18). Non-shockable rhythm typically referred to asystole or pulseless electrical activity (PEA), leading to prolonged ischemia of the brain and other organs, resulting in poor neurological outcomes (19–22).

SOFA score was a simple method for assessing and monitoring organ dysfunction in critically ill patients and had been widely used to assess the severity of critically ill patients and predict their prognosis. The SOFA score comprised six criteria to reflect organ system function (respiratory system, blood system, liver system, cardiovascular system, nervous system, and kidney system) (23). Matsuda et al. (24) conducted a study in 231 cardiovascular arrest patients and found that the SOFA score was lower in those people who survived and SOFA score on admission was a strong predictor of survival at 30 days (OR 0.68, 95% CI: 0.59–0.78). Choi et al. (25) studied the prognosis of 173 patients without-hospital cardiac arrest and found the area under the curve of SOFA score for prediction of 30-day mortality of post-cardiac arrest patients was 0.641 (95% CI: 0.564–0.712). These studies indicated that a decrease in the SOFA score was associated with an increased rate of patient survival, and the SOFA score on admission was significant for predicting the survival rate within 30 days. These findings and our results emphasize the importance of considering the SOFA score in the initial assessment and treatment of patients with cardiac arrest.

When cardiac arrest strikes, time is crucial, with every passing second holding immense significance. Within a span of 4 to 6 min, most patients will start experiencing irreversible brain damage, leading to biological death within a few minutes thereafter. A prompt recognition of cardiac arrest by a bystander, followed by the immediate administration of high-quality CPR, is pivotal in achieving a successful resuscitation outcome. For every 1-minute delay in CPR in cardiac arrest patients, their survival rate decreased by about 10% (26–28). Although numerous studies had demonstrated the importance of bystander CPR when someone witnesses an arrest, the rate of bystander CPR remained relatively low in many countries (29). Nevertheless, we know little about bystanders' perceptions that influenced the decision to start CPR. Our study found that witnessed arrest was independently associated with hospital mortality rate 24 h after CPR, which further reinforced the importance of teaching the early recognition of cardiac arrest.

ROSC was not only a sign of successful CPR but also was the starting point for further treatment and prognosis assessment. In the clinical practice, physicians closely monitored the status of ROSC and adjusted treatment plans accordingly, to improve patient survival rate and quality of life. However, the time to ROSC after CPR varied in different situations, and the ultimate outcomes were also different. In a Swedish cohort study, the median duration of CPR with ROSC was 5 (IQR 2–12) minutes (30). Data from the GWTG-R (Get With the Guidelines-Resuscitation) registry have shown that the median time to achieve ROSC was 11 (IQR 6–21) minutes (31). Although longer duration of resuscitation was associated with worse outcomes, individuals could still benefit from prolonged CPR. Data from the American Heart Association GWTG-R registry showed that 88% of patients achieved sustained ROSC within 30 min (32). These studies indicated that the duration of CPR was closely associated with the success rate of ROSC, but the specific median time to ROSC varied. Therefore, understanding the median time to ROSC can help assess the effectiveness of CPR and may guide clinical decision-making. Our results showed a similar conclusion. The duration from cardiac arrest to the ROSC significantly impacted the survival probability and the shorter duration correlated with improved survival prospects.

Elderly patients were prone to many comorbidities and were therefore more likely to have a poor prognosis after cardiac arrest (33, 34). and some studies had revealed that advanced age was a strong predictor of mortality (31, 35). Sender et al. (36) conducted a study on patients with cardiac arrest in 6 interventional cardiology centers and found that elderly patients over 75 years old were more likely to have adverse neurological function prognosis at 6 months follow-up. Ester et al. (37) retrospectively analyzed 1,285 adult patients with cardiac arrest and found that older patients had worse neurological outcomes and a higher proportion of death after cardiac arrest compared with younger patients. Li et al. (38) retrospectively analyzed the risk factors, and in-hospital outcomes of 320 patients with acute coronary syndrome suffering cardiac arrest and they revealed that age less than 70 was demonstrated to be a strong predictor of survival. However, some studies have found that age was not a risk factor for survival after cardiac arrest, suggesting that when assessing the prognosis of patients who had experienced cardiac arrest, other clinical factors should also be considered in addition to age (39–43).

Alkaline phosphatase (ALP) is a metalloenzyme in membrane-bound glycoprotein, and also an enzyme that catalyzes calcification inhibitor pyrophosphate hydrolysis. ALP is widely distributed in various organs and tissues of the human body, with the highest activity in the liver. The abnormalities can be seen in liver cancer, obstructive jaundice, and other diseases of the liver and bile system, as well as intestinal diseases, metabolic abnormalities, chronic renal insufficiency, and bone metabolism abnormalities (44, 45). Studies have shown that ALP was a new inflammatory mediator of cardiovascular diseases, and the increase of serum ALP level was associated with a variety of atherosclerotic diseases, indicating that ALP was closely related to cardiovascular and cerebrovascular diseases, and its specific mechanism might be leading to increased vascular calcification and vascular dysfunction and thus to promote the release of inflammatory mediators (46–48). Zhong et al. (49) showed that the risk of the early death of stroke patients with high ALP levels was 2.19 times higher than that of stroke patients with low ALP level. Moreover, there was a significant linear relationship between serum ALP level and hospital mortality of cerebral infarction, which further revealed that serum ALP level was associated with poor recovery of neurological function, increased mortality, and overall poor prognosis of cerebral infarction patients. Another study showed that in patients with coronary heart disease, elevated ALP was an independent risk factor for increased 3-year all-cause mortality (50). Therefore, it was speculated that ALP, as an inflammatory indicator, reflected the occurrence and development of cardiac arrest patients to a certain extent, and could be used to predict the prognosis of cardiac arrest patients.

Our nomogram prediction model, based on these selected predictors, had accurate predictive value for hospital mortality in patients who survived 24 h after CPR, which will be beneficial for assisting in identifying high-risk patients and intervention. Compared with the predictive model from Chen et al. (51), we include more variables, up to 54 variables, and thus the predictive model may be more comprehensive and representative.

Our research had some limitations. Firstly, it's an observational study with a relatively small sample size, and selection bias was inevitable, making it difficult to ensure an even distribution of all variables. Secondly, all participants were from Belgium. Further researches were needed to verify whether these findings were also applicable to different ethnicities and external validation data was required to further clarify the accuracy of this model. Thirdly, we analyzed a mixed population of patients with out-of-hospital cardiac arrest and in-hospital cardiac arrest, but there was a significant difference between the two groups, such as cause, prevention, and treatment, and these conditions should be taken into account.

Our research showed that non-shockable rhythm, SOFA score, witnessed arrest, time-to-ROSC, age, and alkaline phosphatase were significantly associated with hospital mortality for these patients who survived 24 h after CPR. Our nomogram prediction model had accurate predictive value for hospital mortality in patients who survived 24 h after CPR, which will be beneficial for assisting in identifying high-risk patients and intervention. Further confirmation of the model's accuracy required external validation data.