The causes of ICU admission directly from the ED clearly varied with both age, sex and type of hospital.
Overall, the finding that intoxication was such a common reason for ICU admission in the adult population stands out. Intoxication caused 46.8% of all ICU admissions in young adults (18–29 years) and 39.4% of the admissions in the age group 30–44 years, which means that a large share of all ICU resources on a national level in Sweden are used to care for intoxicated patients. Yet, overall mortality in intoxicated patients was less than 1 %. Due to lack of data, we could not distinguish between intentional and non-intentional intoxications in the study cohort. According to previous studies on poisoning in adolescents and adults it is, however, reasonable to assume that most of the intoxications were intentional [18, 19].
In the elderly, infectious diseases were the most common reason for ICU admission. This finding stands in contrast to previous European studies, most notably Fassier et al., who reported that respiratory-related diagnoses, and especially acute pulmonary edema, were the most common causes of ICU admission in elderly (≥80 years) patients in France [20]. In addition, in a study by Flaatten et al., respiratory failure was reported as the most common cause of ICU admission in patients ≥80 years old [21]. There may be several reasons for these discrepancies. In the study by Fassier et al., pneumonia was defined as a respiratory condition whereas, in this study, we defined it as an infectious disease. Further, another reason for the differences may be that the studies by Fassier et al. and Flaatten et al. also included patients admitted to the ICU from a ward, not only from the ED, which is likely to alter the spectrum of underlying medical conditions.
In children, the most common causes for intensive care admission from the emergency department were neurological conditions (< 9 years) and intoxications (10–17 years). This finding also differs from a previous study in England and Wales, where respiratory and cardiovascular causes were more common [22]. However, like most earlier studies in the adult population, this study also included patients admitted to ICU from a ward. Other explanations for the discrepancies may be different organization of the health care systems in Sweden and the UK.
The most common cause of ICU admission in males was trauma. According to a previous study from another Scandinavian country, Finland, trauma resulting in intensive care was similarly more common in males than in females [23].
In contrast, intoxication was the most common cause of intensive care admission in females. In fact, the absolute number of males and females receiving intensive care because of intoxication was almost similar, but due to fewer female ICU patients overall, the percentages differ (22.4% of females; 16.7% of males). This result is in line with an earlier study by Lindqvist et al., where ICU-treated intoxication was as common in males as in females [24].
In our study, the panorama of causes for ICU admission as well as the mortality rates differed between hospital types. One reason for this may be that some advanced care is centralized to the academic centers in Sweden [25]. However, this is unlikely to be the sole explanation, since we only included patients who were directly admitted to the ICU from the ED and patients usually visit the ED which is geographically closest to their home. Another probable reason for regional differences could be differences in hospital organization. Larger hospitals often provide intermediate care units for patients in need of continuous monitoring of vital signs but without need for advanced intensive care. In hospitals without intermediate care facilities, often rural hospitals, these patients are typically admitted to the ICU instead. This model of explanation may also be supported by the fact that ICU patients in academic hospitals seem to be sicker than in rural hospitals, as indicated by the SAPS3 value at admission as well as a higher mortality in the ICU.
Deciding on the right level of care for the individual patient is one of the main challenges in the ED. In times of ICU bed shortage, it is even more important to admit the “right” patients to the ICU; those who are too sick to be admitted to a general ward but who could still benefit from ICU care. The COVID-19 pandemic, with a further aggravated shortage of ICU beds, has highlighted the need to optimize ICU bed utilization [26]. However, even in non-pandemic times, ICU bed capacity varies over time and shortages are prevalent in many hospitals [27, 28]. Generally, Sweden has a low ICU bed capacity per capita, compared to other developed countries [29]. Thus, optimization of ICU bed utilization is a constant challenge and identification of ICU patient who would potentially be managed safely at a lower level of care is one potential avenue to reduce ICU strain.
In our study, 12.5% of the patients were admitted to the ICU for observation. This was most common in intoxicated patients, of whom one third were admitted to an ICU mainly for observation. Further, intoxicated patients had exceptionally low mortality (0.3%) compared to other groups of ICU patients. Taken together, these findings raise the question whether a large share of intoxicated patients admitted to the ICU from the ED really benefit from ICU care, or if they could instead be as safely managed in intermediate care wards or on general wards provided that the staff is trained accordingly, and relevant standard operating procedures are in place. To answer this question, an important first step will be to develop tools to help identify patients with a high risk for adverse events among intoxicated ED patients considered for ICU admission.
The present study, being a retrospective, register-based investigation, has some general limitations. Firstly, coding of diagnoses is done at ICU discharge, which entails a risk of recall bias. However, the coding is typically done by the attending intensive care physician who has been responsible for the individual patient, which should reduce the risk of incorrect coding. Further, the Swedish Intensive Care registry is validated for logical errors, but there is no detailed validation of the coding. There is no linkage between the coding in the Swedish Intensive care Registry and economic compensation, either to the hospital or the individual physician. Thus, it is unlikely that coding is influenced by economic incentives.
Since the present study is based only on Swedish data, we are unable to determine the generalizability of our findings to other countries and healthcare systems. Most likely, the panorama of ICU discharge diagnoses will vary according to the overall panorama of disease in the population, as well as factors associated with the organization of the healthcare system. This warrants for similar studies in other countries.
In summary, the most common ICU discharge codes for patients admitted from the ED were intoxication, trauma and neurological conditions. However, ICU admissions from the ED vary substantially with age, sex and hospital type.