In this national cross-sectional study at Swedish EDs during 24 h, we provide a snapshot of current Swedish ED crowding, boarding and staffing, which has never been done before. We observed that boarding was common and occupancy rates were generally high, primarily in academic EDs but also in urban EDs. On average, patient to staff ratios for nurses were on par with internationally reported levels (see below), but lower for physicians. There were more patients per staff at academic centres compared to rural hospitals. Workload was mostly perceived as low to moderate, which indicated limited staff problems related to crowding during the study period.
Occupancy rate correlated modestly with workload, which suggests that these may reflect different aspects of crowding. Workload was subjectively assessed by a single senior provider at each ED which limits the generalizability. However, subjective provider judgement was used as an outcome measure in the original NEDOCS trial and this has been validated in several different settings [19, 28, 29]. Physicians’ judgment has also proved to be equal or superior to structured decision support tools in many types of clinical decision-making ranging from imaging in trauma to the investigation in suspected pulmonary embolism [30].
Boarding was prevalent at many sites during this study. Generally however, boarding was reported as lower compared to the limited data from the United States (US) and Australia published so far. In a US cross sectional study of 89 EDs, 22% reported boarding patients and 73% of EDs had more than 2 patients boarding [9]. In a registry study of 139 509 ED visits in the US, median boarding time was 79 minutes [31]. In a study of 72 EDs in Australia, boarding ranged from 2 to 22 patients at two time points [10]. The difference in findings between the present and previous studies may be due to sampling errors or temporal effects, but it may also reflect possible differences in health care systems. Lack of inpatient beds is usually the basis for boarding patients in the ED. Since Sweden has fewer inpatient beds per capita than the US and Australia, our results support the claim that boarding may not be directly related to the number of hospital beds, but also to resource utilisation [32], both in single hospitals and in the system as a whole. It is important to note that our definition of boarding did not include a minimum waiting time after the decision to admit, and that we did not gather any further information regarding the admissions.
The staffing ratios were comparable at all study sites with most variation observed around midnight. This finding likely reflects that staffing is reduced at night-time and that staffing ratios therefore become more dependent on the inflow of patients. We did not collect information about work shifts at each ED and cannot exclude that this may explain some of the variation in staff ratios. The emergency medicine literature provides little data for comparison, but Schneider et al. reported higher mean ratios for nurses (4.2) and physicians (9.7) in the US in 2003 [9]. The difference, particularly for physicians, may partly be due to different denominators since we registered all physicians irrespective of training level in this study. In Sweden, a majority of the current ED physicians are pre-interns, interns or residents and only a minority are on site consultants [14]. This may result in higher numbers of physicians working in the ED compared to the US, where EDs are primarily staffed by residents and consultants. There are no national recommendations for staffing ratios in Sweden but our results are within the four patients to one nurse ratio legislated in the US state of California [33].
During the 24 h period, four study sites (11%) noted disturbances in the EHR or support systems, and this has previously been associated with increased ED crowding [34]. All EDs in Sweden use EHRs with a range of digital support systems for radiology, laboratory and other ancillary facilities. The reports may thus be an indicator of the fragility of complex digital systems to which ED providers must adapt. The lack of adverse events suggests mature systems with some resilience against unexpected downtime, leading to no serious disruption of clinical work. However, further studies will be needed to determine the frequency of EHR disturbances and their effects on emergency care.
Limitations
This was a cross sectional study during only 24 h, and the generalisability of the results is therefore limited. There may be both seasonal differences in the demand and availability of healthcare resources. However, given the range of EDs both in size and geographic location, we believe that the results are a representative snapshot of the ED situation on a national level in Sweden.
The response rate was 51% among the eligible EDs regarding patient and crowding data, which is quite high compared to similar studies. Again, generalisability was most likely increased by the wide range of ED size and location. However, the fact that so many EDs chose to not participate emphasizes the need for mandatory and public reporting of this type of information for all EDs.