Article Text
PDF
Abstract
Objective Many believe that hospital crowding manifesting in the ED with the boarding of admitted patients is a result of significant numbers of acute hospital beds being occupied by patients awaiting discharge to nursing homes, step-down facilities or home with or without additional support. This observational study was performed to establish the actual relationship between boarders in the ED and patients experiencing delayed discharge.
Methods Data relating to the number of patients in the ED and their points in their patient pathway were entered into a logbook on a daily basis by the most senior doctor on duty. 630 days of observations of patients boarded in the ED were compared with the number of inpatients with delayed discharges, obtained from the hospital information system, to see if large numbers of inpatients with delayed discharges are associated with crowding in the ED.
Results Two years of data showed an annual ED census of more than 47 000, with a daily mean ED admission rate of 29.85 patients and a daily mean ED boarding figure of 29 patients. A mean of 15.4% of the 823 hospital beds was occupied by patients with delayed discharges, and the hospital ran at, or near, full capacity (99%–105%) all the time. Results obtained highlighted a statistically significant relationship between delayed discharges in the hospital and ED crowding as a result of boarders (p value<0.001, with a regression coefficient of 0.16, 95% CI 0.12 to 0.20). The study also showed that the number of boarders was related to the number of ED admissions in the preceding 24 hours (p=0.036, with a regression coefficient of 0.14, 95% CI 0.05 to 0.28).
Conclusions Delayed hospital discharges significantly contribute to crowding in the ED. Healthcare systems should target timely discharge of inpatients experiencing delayed discharge in an urgent and efficient manner to improve timely access to acute hospital beds for patients requiring emergency admission.
- emergency department
- crowding
Statistics from Altmetric.com
Key messages
What is already known on this subject?
Crowding in the ED is a challenge faced by both patients and healthcare personnel and has been shown to impact negatively on ED functioning. A number of factors including high hospital occupancy have been identified as contributing significantly to crowding in the ED.
What might this study add?
Our study assesses and shows an association between the component of hospital occupancy relating to inpatients with delayed discharges and crowding in the ED. This finding emphasises the need to optimise and resource discharge planning.
Introduction
Crowding in the EDs exists where the number of patients waiting to be seen, undergoing assessment and treatment or waiting for departure exceeds the physical or staffing capacity of the department, and it has been shown to impact negatively on ED functioning.1 Crowding which is more descriptive than the commonly used term ‘overcrowding’2 is said to be a very grave but solvable issue in the hospital system.3 Once considered an ED phenomenon, ED crowding is now understood to be a system-wide problem which receives much attention in the lay and medical press. Widespread ED crowding has been cited by multiple studies in the literature.4–7 Ninety-one per cent of hospital ED directors report crowding as a problem.8
A key contributor to crowding in the ED is the number of ‘boarders’, which is a term used to describe admitted patients who are housed in the ED while waiting for inpatient beds to become available.9 In the same way that patients experiencing delayed discharge no longer require care in an acute hospital, boarded admitted patients in the ED no longer require the expertise of Emergency Medical personnel and should be on the appropriate ward to address their ongoing needs. EDs are designed for rapid triage, stabilisation and initial treatment. When boarding in the ED causes ‘gridlock’, the ED becomes the site for ongoing (ie, longitudinal) care in the acute phase of hospitalisation.10 Admitted patients waiting in the ED for inpatient beds for more than a day have, additionally, been shown to endure 10%–20% longer hospital stays (with higher attendant costs) than those patients who can be moved up to the wards more quickly.11
The inability of patients requiring admission through the ED to gain access to appropriate hospital beds within a reasonable amount of time is referred to as ‘access block’.1
Many factors have been linked to causing access block, one of which is high occupancy rates of hospital beds. Foster et al established in their study that ED length of stay was significantly associated with hospital occupancy. They found that a 10% increase in hospital occupancy increased the ED length of stay by 18 min.12 Bagust et al13 highlighted in their study more than 15 years ago that acute hospitals that operate at bed occupancy levels of 90% or more face regular bed crises, with the associated risks to patients.
In acute hospitals, some beds are occupied by long-stay/alternate level of care patients who no longer require acute hospital care and who would be best served by transfer to a long-term facility or to home with increased supports in place, and these patients are sometimes pejoratively referred to as ‘bed blockers’.14 There are several studies in the literature that correlate delayed discharges from hospitals with crowding in EDs,14 ,15 but to our knowledge, this is the first study that determines the actual statistical association. This observational study was performed to determine if there is a quantifiable association between the number of patients experiencing delayed discharge from the hospital and the number of ED boarders.
Methods
The ED in which the research took place provides care to approximately 50 000 adult patient attendances and is situated in an 823-bed urban teaching hospital. The ED is among the most crowded in Ireland and serves a large population of North Dublin and surrounding counties. The hospital also serves as a neurosurgical, nephrology and oncology centre for a major part of the country and is the national centre for renal transplantation.
The information technology systems in the hospital in which the study took place were installed over 20 years ago. The hospital information services keep records of the data of patients' flow in the hospital, including the number of patients admitted and discharged to and from the hospital, both on an emergency and on an elective basis. It also keeps records of delayed discharges from the hospital, which is updated on a weekly basis. The business intelligence unit of the hospital defines a long-stay inpatient or a delayed discharge as ‘a patient who remains in hospital after a senior doctor (consultant or registrar grade) has documented in the medical chart that the patient can be discharged’. The reasons for delay are grouped into three main areas: those relating to patients going home, patients going to long-term nursing care and other (patients who require some other forms of service before it can be decided if they can go home, eg, rehabilitation). The data for delayed discharges for the calendar years 2010 and 2011 were obtained from the hospital information system and entered into an excel spreadsheet (Microsoft Office 2010).
Prospective data were collected for 2010 and 2011 in the ED, where a senior ED doctor on duty, each morning at 08:00, noted the distribution of patients and their points in their pathway of care in ED and manually entered the data into a logbook. The time points included those waiting, those being assessed by the ED staff, those awaiting or undergoing assessment by the on-call teams and those admitted patients boarded in the ED, pending the availability of a ward bed. Distinctions were made between weekday and weekend attendances (which also included public holidays). These data were then merged with the data of patients experiencing delayed discharge from the hospital.
We divided the cohort of patients experiencing delayed discharge into two groups, a low and a high group, based on the weekly median number of patients with delayed discharge. We then calculated the power of the study based on the daily mean number of ED boarders in each group, which was 27.2 (SD 9) for the low group and 29.9 (SD 10.7) for the high group. This suggested that a total number of 424 observations will give a power of 80% to detect a difference at the 5% level of significance.
Distributional properties of skewness and kurtosis for both the dependent variable (ED boarder numbers) and independent variables (total ED admissions, weekend/holiday effect and delayed discharges) were assessed. With information derived from these tests, delayed discharge numbers in particular were deemed to be non-normally distributed and were analysed using median, IQRs and quantile regression. Quantile regression uses the median coefficient of the independent variables rather than the mean used in standard linear regression. Medians are far more robust to skewed data that were detected in the delayed discharges. Medians and quantile regression were also used to examine the relationship between boarder numbers and hospital admission rates. Mean and SD were used for daily ED attendances, daily admissions from ED and daily ED boarders. Weekend dates and public holiday dates were identified and grouped versus weekday dates to form a binary variable to be assessed as a potential confounder. Statistical software, Stata, was used for the analysis (V.13 SE, College Station, Texas, USA). A type 1 error of 5% (p<0.05) was deemed to be significant.
Results
A total of 43 815 patients were admitted, both electively and on an emergency basis, to the hospital in the 2-year study period, with 21 722 admitted in 2010 and 22 093 in 2011. There was a total of 456 095 bed days used for 2010 and 2011. The mean length of stay was 11.0 days in 2010 and 10.1 days in 2011. The number of ED attendances of patients, both new and return, for 2010 was 47 168, and for 2011 it was 49 342. The mean daily number of ED attendances was 132 patients for the 2 years, with a mean of 30 patients being admitted in each 24-hour period. The mean number of patients boarded in the ED daily at 08:00 was 29 patients (range 1–59) (table 1).
Hospital figures for the 2 years
The median number of delayed discharges in the hospital for the year 2010 on a weekly basis was 110 (range 80–150), and for 2011, the median was 87 (range 76–108). The cause of delayed discharge for these patients was also assessed. They were broadly identified as patients awaiting community services, awaiting housing or adaptation to their home, awaiting funding for approved homecare packages, awaiting external rehabilitation, awaiting hospice care, being a ward of court, awaiting availability of a nursing home following approved subvention, pending work in progress for a homecare package and nursing home subvention work in progress, patients or families requesting publicly funded long-term care beds and patients that require residential care due to high medical care needs. We found that the highest numbers of patients with delayed discharges were the ones that had requested publicly funded long-term care beds, with a mean of 86 patients per week in this category. This was followed by patients requiring residential care due to high medical care needs with a mean of 29 patients per week and those awaiting external rehabilitation with a mean of 26 patients per week. Most of the patients with delayed discharge were over 65 years of age, with a mean of 110 and 79.1 patients per week in 2010 and 2011, respectively, whereas the mean number of patients with delayed discharge per week under the age of 65 years was 18 and 11.2 in 2010 and 2011, respectively.
Data were missing in relation to boarders in the ED for 97 days, which over a 2-year period represent 13.3%. There were 5 days of missing delayed discharge data. The number of missing days for both ED boarders and delayed discharge was 100. This represented 13.7% of the data. It was decided to model the available data using 630 observations.
Regression analysis of the number of patients boarded in the ED and the number of delayed discharges shows a significant association (p<0.001, with a regression coefficient of 0.16, 95% CI 0.12 to 0.20) (figure 1). For each additional long-stay patient, there is a predicted increase of 0.16 in the number of boarders in the ED. The results for 2010 and 2011 for patients with delayed discharge and boarders in the ED are displayed in figure 2. When the patient numbers with delayed discharge are arranged in ascending order, the commensurate boarder numbers demonstrate significant association (figure 3), that is, as delayed discharges increase, patients boarded in the ED increase. The number of boarders was related to the number of ED admissions over the preceding 24 hours (p=0.036, with a regression coefficient of 0.14, 95% CI 0.05 to 0.28). These results are presented for the 2-year period in figure 4 and were adjusted for weekend and public holiday confounder effects.
Scatter plot of mean daily delayed discharge patients and number of ED boarders.
Graph depicting the number of ED boarders at 08:00 every morning and the number of delayed discharges for the years 2010 and 2011.
Delayed discharges in ascending order and the corresponding number of boarders in ED.
Graph depicting the number of ED boarders at 08:00 every morning and the number of admissions over the last 24-hour period.
Discussion
ED crowding is an international problem and should be considered a public health threat. When compared with the US teaching hospitals who have a mean ED boarding period of 3.5 hours (median 2 hours), the institution in which this research took place has a previously published mean of 16.1 hours waiting for admission following a bed request.16
While previous studies have suggested that the use of acute care beds by patients under suitable for alternate level care contributes to the problem of ED crowding by preventing the admission of emergency patients to hospital beds, this study has shown a statistical correlation between the number of patients with delayed discharges and the number of boarders in ED. In a study in Ontario, Canada, patients with delayed discharges occupied an average of 10% of total staffed bed capacity.14 Some individual hospitals reported the percentage to be much higher, at 20%–25%.14 The hospital in which our study took place had percentage occupancy of 99.9%–105.1%, and the mean percentage of patients with delayed discharges was found to be 15.4%.
The Irish Health Forum Steering Group (including the Health Service Executive, Department of Health and Children, Department of Finance, the European Social Research Institute, Central Statistics Office and clinicians) published a report in 2008, highlighting the fact that in Ireland, 37% of total inpatient beds are occupied by patients no longer expected to be in hospital. A patient typically spends between 0.6 and 1.9 days longer in an Irish hospital than they would in a UK hospital for the same treatment.17
Although our study is a single-centre study, the hospital is one of the worst affected by the crisis of crowding in Ireland, and as such it is well placed to examine variables related to this serious issue.
This study highlights the need to facilitate the timely discharges of patients that have completed their acute course of admission. The discharge process should start at the admission point, as it is the mismatch between demand and supply of beds that promotes delays and bottlenecks in the system.18
Many factors have been identified that lead to delayed discharges. Some of them are a reduction in the number of beds in nursing homes; problems in funding from social service budgets; waits for assessments from therapists or social services; waits for community services or for equipment to be ordered, delivered and installed; reduced junior doctor hours and introduction of shift work, resulting in less time on the wards to see patients and their relatives, leading to inefficient communication and poor operation of discharge procedures.19 Proposed solutions include good management of available resources as well as instituting new changes to enhance the quality of service and care to the patients. In the past few years, many trusts in the UK have appointed discharge facilitators/coordinators, who are involved in targeting delayed discharges and finding and expediting discharges during periods of pressure at the admissions end. Many have also set up discharge lounges. There is also great scope to reduce length of stay by increasing the frequency of ward rounds and considering the probable content and timing of a patient's discharge package earlier, potentially at the time of admission.19 The roll-out of primary care teams in the community, the resourcing and development of services for an older and a more dependent population, chronic disease management and knowledge from information systems being used to examine data and execute planning accordingly are other possible solutions.20 Long-term planning is required to meet the demand for non-acute care facilities. There is a need to define the catchment areas for each local community in terms of demographic characteristics, finding available community services and establishing the requirement for creating new facilities and services.
Study limitations include the fact that the study was performed in a single, although profoundly crowded, centre and that there were some missing data. Over the period of 2 years, we found missing data of 13.7% for boarders in the ED and delayed discharges from the hospital. Data were found to be missing randomly; therefore, the missing data were deemed unbiased. The missing data often related to a failure to log the patients in the ED at 08:00 by a member of the senior emergency medicine team. It was decided to model the available data using 630 observations, which showed the association between boarders and delayed discharges with a regression coefficient of 0.163 (p value <0.0001, CI 0.12 to 0.20). Imputing the missing ED data by predicting results from the regression model and then proceeding to a model with a full dataset showed very similar results with a regression coefficient of 0.168 (p value<0.0001, CI 0.15 to 0.19). This calculated association gives a statistical assumption that we can expect 30 boarders in the ED when the number of delayed discharges in the hospital reaches 103.
Though this study confirms an association between delayed discharges from the hospital and boarding in the ED, this correlation is not as strong as might have been expected. This is reflective of the fact that while a mean of 15.4% of the hospital bed base is occupied with patients experiencing a delayed discharge, the hospital runs at, or near, full capacity (99.9%–105.1% occupancy) almost all of the time. Whether a hospital bed is occupied by a patient with delayed discharge, an emergency admission or a planned admission, if it is occupied, it is not available for those requiring admission from the ED, and as a result will mean that the next patient requiring emergency admission will be boarded in the ED. Boarding can only be avoided by having available ward beds.
The correlation between the previous day’s patients who required admission through the ED and the number of admitted patients boarded in the crowded ED shown in our study is indicative of the fact that in the hospital where this research took place, the majority of patients requiring emergency admission wait until the next day for a hospital bed to be made available, while the hospital has in excess of 15% of its bed base occupied by patients who no longer need to be in an acute hospital.
Conclusion
This study confirms the commonly held assumption that delayed discharges in the hospital are associated with access block and increased boarding of patients in the ED, which contribute to crowding. Patients who would be better served in alternative facilities should not be kept in acute hospital beds that are required for those needing emergency hospitalisation, as doing so contributes to dangerous ED crowding which compromises emergency care delivery.
References
Footnotes
Contributors PG developed the research question and methods, was involved in the study design and advised on the research. FM, DO, PG, EO'H, CL, SK, AH and PH gathered the data. PO’K provided statistical analysis and advice on study design and performance. FM, PG, PO’K and DO were involved in data analysis and wrote the paper.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.