Maximum emergency department overcrowding is correlated with occurrence of unexpected cardiac arrest

Emergency department (ED) overcrowding is a global phenomenon with complex causes that are not fully understood [1, 2]. The imbalance of supply and demand in ED capacity can result in hazardous effects both on physicians and patients [3], and previous studies have found that overcrowded ED conditions increased physician decision-making time [4], led to more medication errors [5], decreased quality of care [6], and increased mortality during hospital stay [7, 8]. These potential dangers can be particularly detrimental for patients who needed time-critical interventions to effectively treat emergency conditions such as acute stroke [9], acute respiratory failure [10], and septic shock [11].

In-hospital cardiac arrest is a critical situation requiring prompt intervention and continues to be a major public health burden [12, 13]. A previous study found that the quality of cardiopulmonary resuscitation (CPR) of out-of-hospital cardiac arrest (OHCA) was not associated with overcrowding because emergency medical services could alert the ED CPR team of incoming OHCA patients, allowing time for the team to activate [14]. Compared to OHCA, many in-hospital cardiac arrests (IHCAs), especially during ED stay, are unexpected but considered to be preventable if early identification of at-risk patients and adequate interventions are possible [15]. However, there is limited evidence as to whether ED crowding influences the occurrence or outcomes of IHCA. We hypothesized that ED overcrowding correlates with IHCA occurrence. To test this hypothesis, we used a prospective registry of in-hospital cardiac arrest to evaluate the association between the ED occupancy rate in various timing and unexpected cardiac arrest.

During the study period, 629 adult, non-traumatic patients experiencing OHCA and IHCA were enrolled in our registry. Of these, 187 patients experienced IHCA and 108 patients survived.

The demographic and clinical details of the enrolled patients are summarized in Table 1. The median age was 67.0 years, and patients were predominantly male in both groups. Hypertension (35.3%) was the most common past medical illness in both groups. The survivor group had a more frequent previously diagnosed myocardial infarction (10.2 vs. 1.3%), arrhythmia (20.4 vs. 7.6%), heart failure (16.7 vs. 7.6%), and diabetes (38.0 vs. 19.0%), but less malignancy (25.0 vs. 40.5%) than that of non-survivor. The presumed cardiac source was the leading cause of arrest, and respiratory origin was more common in survivor than in non-survivor (27.8 vs. 15.2%). Additional table represented the total IHCA distribution according to the time and showed no definite patterns (Additional Table 1).

Table 2 shows the variables associated with cardiac arrest. Among 187 total IHCA patients, 125 patients (66.8) were successfully ROSC after CPCR. Non-survivor had a higher proportion of DNR after ROSC (36.7 vs. 7.5%). All IHCA events were witnessed, and the median no-flow times were 0 min (IQR 0.0–0.0 min). Pulseless electrical activity was predominant in both groups, and initial ventricular fibrillation rhythm was more common in patients with survivor than with non-survivor (13.0 vs. 2.5%). Regarding the treatment-related factors, the survivor group had more hypothermia, percutaneous coronary intervention, and extracorporeal membrane oxygenation than the non-survivor. Time to first defibrillation, epinephrine injection, and endotracheal intubation was similar in both groups, except for the duration of CPR (6.0 vs. 27.0 min; survivor vs. non-survivor, respectively). Overall survival discharge rate was 24.6%, and 20.3% of patients showed favorable neurologic outcomes at discharge. All ED occupancy rates, including presentation, arrest, maximum, and average, were similar in both groups.

Figure 1 presents the frequency of cardiac arrest according to time from ED arrival. About two-third cases occurred and the median time was 72.0 min. Figure 2 shows the correlation between the occurrence of cardiac arrest patients and ED occupancy at specific times for each ED zone. Moreover, Fig. 3 represents the association between the death in ED and the maximum occupancy rate. In contrast to the occurrence, mortality in ED was not correlated with the ED occupancy rate.

IHCA occurrence was positively correlated with most of ED occupancy rate except occupancy rate at ED presentation and average during ED stay of the urgent zone (Table 3). All ED occupancy rate in the critical zone was positively correlated with IHCA occurrence, and maximum occupancy rate of IHCA was the most strongly correlated (Spearman’s ρ = 1, P < .01). Maximum ED occupancy of the urgent zone also yielded a good positive relationship with IHCA occurrence. In the entire ED, all four occupancy rates including presentation, arrest, maximum, and average had strong positive correlations with IHCA occurrence. Table 3 summarizes the results of Spearman rank correlation tests for the relationship between cardiac arrest occurrence and ED occupancy.

In this study, we compared various types of ED occupancy rate at presentation, at arrest, maximum, and average during ED stay for IHCA occurrence. We found that the maximum ED occupancy rate is most strongly associated with IHCA occurrence. The correlation between ED occupancy rate of the critical zone and IHCA occurrence was stronger than that of the urgent zone. Meanwhile, any ED occupancy rate did not show the relationship with the outcomes of IHCA.

We found that higher maximum ED occupancy rates, which may indicate the most hazardous and chaotic times in the ED, were strongly associated with sudden IHCA occurrence. In a recent observational retrospective study conducted by Chang et al., ED crowding was roughly associated with IHCA occurrence [23]. In this study, ED bed occupancy rate (EDBOR) was used as an overcrowding index by calculating the ratio of the total number of filled beds divided by the total number of licensed ED beds. However, EDBOR is an index that has never been used before and it is difficult to compare with other studies. Moreover, the EDBOR might lead to substantial selection bias because filled beds may not reflect the proportion of critically ill patients at risk of experiencing IHCA in the ED, such as patients waiting for a bed. Ye et al. recently found that there was no association between IHCA events and EDBOR when they calculated the IHCA incidence rate [24].

We also questioned the value of only measuring ED occupancy rate only at ED presentations, which is done in most ED occupancy rate studies, as it might not be sufficient to evaluate the types of overcrowding that ED patients may experience during their ED stay. Other measurements, such as at the time of presentation, arrest, or maximum occupancy, or average ED occupancy rate during ED stay may be better parameters for reflecting overcrowding. To the best of our knowledge, this study is the first study that compared the relationship between IHCA occurrence and ED occupancy rate at these times. Chang et al.’s study used three arbitrary cut-off values of the bed occupancy rate [23], but ED occupancy rate crowding measures should be dealt with as a continuous variable because there is no standard threshold. In our study, we normalized with IHCA incidence per hour for each ED occupancy rate interval (0.2 increment for the critical zone and 0.4 increment for the urgent zone) and found there was a strong association between unexpected IHCA and ED occupancy rate. Maximum ED occupancy might imply the busiest or worst crowding and caused IHCA through multifactorial mechanisms, including decrease surveillance for at-risk patients, decrease medical resources to initial resuscitation, increase turnaround time of laboratory and image workup, or increase medical errors. Meanwhile, our study also proved that not only overall crowding but the portion of critically ill patients dramatically impacts on the occurrence of IHCA. In the Spearman rank correlation test, all rho values of the critical zone were higher than those of the urgent zone. Moreover, the critical zone showed a definite positive correlation with IHCA occurrence at sub-maximal level of ED occupancy rate (≤ 0.9). These trends would imply that even a small number of critical patients could compromise the quality of ED services and surveillance ability of at-risk patients.

We also found that although higher ED occupancy rates showed a positive relationship with the occurrence of unpredicted CA, it was not correlated with ED mortality. Multifactorial reasons could explain this. One of the main reasons was that whenever IHCAs occurred, the available number of medical personnel involved in CPR regardless of the degree of the crowding. This result also could imply that monitoring and trying to control ED occupancy rates could reduce the incidence of IHCA. Based on the ED occupancy rate, the study facility has tried to diminish overcrowding through performing ambulance diversion, shortening the required time for decision-making, and transferring the patients to nearby hospitals whenever there were no available beds in general wards or ICUs to reduce the number of patients waiting for boarding. Similar results were reported in the previous study although they focused on OHCA [14].

Our study had several limitations. First, the retrospective cohort design imposes intrinsic limitations on the study, making it difficult to generalize the findings. Second, we could not exclude various confounding factors to develop cardiac arrest, such as the severity of illnesses before the occurrence of cardiac arrest. In the past, South Korea used the modified Canadian Triage and Acuity Scale for triage; however, the Korean Society of Emergency Medicine reformed this scale and introduced the revised scale called the Korean Triage and Acuity Scale since 2016 [16]. Therefore, our data had no consistent triage scale because the system was changed in the study period. Third, the registry had information only about enrolled patients experiencing IHCA. Therefore, we could not analyze and evaluate the risk factors of IHCA in overcrowded ED. Lastly, this study included a relatively small number of patients at a single center.

We found that the maximum ED occupancy rate is strongly associated with the IHCA occurrence. Adequate monitoring and active controlling of the maximum occupancy rate would be important to reduce unpredicted cardiac arrest in ED.