Daily output measures of Emergency Department in association with seasonality and day of the week: A retrospective cohort observational study

Abstract

To evaluate the daily output measures of the emergency department (ED) in association with seasonality and the day of the week. A retrospective cohort of ED visits to tertiary medical centers between 2016 and 2020. The research unit was each day during the study period. The independent variables were season and day of the week. The dependent variables were ED visits, admission and dropout rates, and duration of ED discharge. The comparison of means was evaluated using ANOVA. Statistical significance was set at P < .05. There were 1826 days, 792 thousand visits, 58% were female. Admission rate 28%, duration to discharge 3.8 h, dropout rate 2%. The average daily visits by season ranged from 101% of the overall average in autumn to 97% in spring. Average daily visits by day of the week were significantly different, with the highest on Sunday (Israel’s first working day of the week), 124% of the overall daily average, and the lowest on Saturday (weekly day off) with 70%. Saturdays had the highest admission rate of 30% and 28% of the overall rate. There was a moderate dependency between the ED duration and discharge, with a dropout rate of r² = 0.19. The average daily visits were not affected by season but differed considerably by day of the week. Admission rates varied slightly by season but were similar by day of the week apart from Saturdays. This may be attributable to the case mix on Saturdays or less restriction to admit when the number of visits is low. We recommended each Emergency Department to evaluate its daily output measures dependency with seasonality and day of the week for operational optimization.

1. Introduction

The number of emergency department (ED) visits has increased steadily, even beyond natural growth.^[1] This may be attributed to aging demographics, as the elderly are likely to have complex health conditions and accordingly consume urgent care.^[1] During the last decades, it has been established that certain conditions in which the window of opportunities for successful intervention is narrow should be referred to as undelayed directly to the ED (e.g., suspected myocardial infarction or ischemic stroke).^[2,3] Another reason may be increasing expectations regarding timely accessibility to healthcare that, when unmet, results in over-utilization of ED. Although it is believed that improving access to primary care and community consulting services may reduce ED utilization, no alternative can fully replace ED. The primary role of the ED is the availability of immediate emergency services. In the past, the expected outcome of ED was limited to either admission or discharge, but during the last decades, the expectation is that ED provides comprehensive clinical evaluation, diagnosis, and even urgent interventions.

Seasonality has been associated with morbidity and mortality and is assumed to affect ED visits. The day of the week has already been proven to be the main predictor of visits.^[4] High volume may affect other output measures, such as admission rate, discharge rate, dropout rate, and duration of discharge.^[5] Overcrowded ED can impair accessibility in a timely manner. Dropout rates vary among populations and may reflect diverse expectations, tolerability, and dissatisfaction.^[6]

Policymakers and insurers are concerned about increased ED use, as they are associated with relatively higher costs than non-ED alternatives.^[4]

This study aimed to evaluate the association between daily output measures in ED with seasonality and with a day of the week.

2. Methods

This retrospective cohort is based on a single tertiary medical center database. The inclusion criteria were emergency department visits between January 1, 2016, and December 31, 2020. The facility is tertiary acute adults’ 1000 beds hospital, located in the suburbs of Tel Aviv metropolis, Israel. An independent pediatric hospital is adjacent; hence, pediatric visits are episodic. There is an entirely separate gynecology and maternity ED that were not included in this work.

The research unit was each day during the study period. The visit date was defined according to the ED registration. Seasonality and the day of the week were derived from the visit date. All output measures were recorded on the visit date. The daily admission rate was defined as the daily inpatient admissions out of the overall daily ED visits. The discharge rate was defined as the daily discharge out of the overall daily ED visits. The dropout rate was defined as the daily number of dropouts from the overall daily ED visits (regardless of the process it had occurred). The duration to discharge was defined as the average daily discharge duration. The discharge rate is practically complementary to the admission rate; hence, it is not presented.

Ethical approval was waived as it was considered unnecessary given the nature of the study that had utilized cumulative administrative output data attributable to each date rather than an individual patient.

3. Analysis

All analyses were conducted using SPSS (IBM, Chicago, USA) version 26. Comparison of means by sex, season, and day of the week was evaluated using analysis of variance (ANOVA). Associations, including categorical variables such as seasons or day of the week, were evaluated using the Spearman correlation coefficient. The association between the numerical variables was evaluated using Pearson’s correlation coefficient. Predefined association categories according to correlation coefficient:

r² < 0.05 unassociated

0.05 < r² <0.10 mild association

0.11 < r² < 0.20 moderate association

0.21 < r² < 0.50 considerable association

r² > 0.5 strong association

P < .05 was considered statistically significant.

4. Results

During the study period, there were 1826 days and 791,853 ED visits, among which 224,471 were inpatient admissions (28% of all visits), 560,740 were discharged (70% of all visits), and 16,671 (2.1% of all visits). Females accounted for 58% of all visits.

Figure 1 presents daily average visits. Figure 1A presents the daily average visits by season, which is insignificant, although statistically significant differences ranged from 97% of the overall daily average visits in spring (the lowest seasonal average) to 103% of the overall daily average in summer (the highest seasonal average). Figure 1B presents the daily average visits by day of the week. Sunday, Israel’s first working day after the weekend (similar to Monday in the western world) had the highest daily visits, 124% of the overall daily visits average, and Saturday, Israel’s weekend day off (similar to Sunday in the western world) had the lowest daily visit load of 70% of the overall daily visit average.

Figure 1.

Daily average visits and 95% confidence interval. (A) By season; (B) by day of the week.

There was a mild statistically significant difference between the sexes’ admission rates of 28.7% in women and 28.2% in men.

Figure 2 presents admission rates. Figure 2A shows diverse admission rates by season ranging from 27.5% in summer to nearly 30% in winter. Figure 2B shows diverse admission rates by day of the week, in which Saturday is exceptionally high at 30%, while all other days (Sunday–Friday) ranged from 27.8% to −28.7%.

Figure 2.

Admission rates and 95% confidence interval. (A) By season; (B) by day of the week.

Duration to discharge was statistically significant longer in males 3.93 h vs. 3.73 h.

Figure 3 presents the duration to discharge. Figure 3A shows diverse duration to discharge by season, the longer duration in winter at 3.92 h (with an extremely wide confidence interval) and the shortest in summer 3.78 h (not very different than autumn and spring). Figure 3B presents diverse duration to discharge by day of the week, the highest on Sunday 4.05 h (vs. 3.83 h overall duration average), and the lowest in Saturday 3.46 h slightly shorter than Friday 3.65 h (in which the confidence interval was very wide).

Figure 3.

Daily average duration to discharge and 95% confidence interval. (A) By season; (B) by day of the week.

There was a statistically significant higher dropout rate among men (2.3% vs. 1.9% among women).

Figure 4 presents dropout rates. Figure 4A shows no considerable differences in dropout rate by season, the highest in autumn 2.19% and the lowest in winter (2.03%). Figure 4B presents a diverse dropout rate by day of the week, the highest on Sundays (2.3% vs. 2.1% overall) and the lowest on Saturdays (1.9%).

Figure 4.

Dropout rates and 95% confidence interval. (A) by season; (B) by day of the week.

Table 1 presents the association between days of the week by Spearman correlation coefficient with output measures, which showed a strong dependency on ED visits (r² = −0.54), mild dependency with admission rate (r² = 0.06), no dependency with dropout rate (r² = −0.04), and mild association with duration to discharge (r² = 0.1).

Table 1.

Spearman correlation between day of the week with emergency department (ED) output measures.

	Correlations	Visits	Daily admission rate	Daily duration to discharge	Daily dropout rate
Day of the week	Spearman	−0.538	0.063	−0.15	−0.038
	Sig. (2-tailed)	<0.01	<0.01	<0.01	0.022
	N	1826	1826	1826	1826

Table 2 presents the association between daily visits by Pearson correlation coefficient and other ED daily output measures; there was moderate dependency on admission rate (r² = 0.14), moderate dependency on daily duration to discharge (r² = 0.11), and no dependency on daily dropout rate (r² = 0.023).

Table 2.

Pearson correlation between emergency department (ED) visits load and other output measures.

	Correlations	Daily admission rate	Daily duration to discharge	Daily dropout rate
Visits	Pearson	−0.14	0.11	0.023
	Sig. (2-tailed)	<0.01	<0.01	0.17
	N	1826	1826	1826

There was a moderate dependency between daily dropout rates by Pearson correlation coefficient with daily duration to discharge (r² = 0.19, N = 1826, P < .01).

5. Discussion

ED visits were relatively similar by season but showed considerable diversity by week. Sundays had the highest number of visits, and Saturdays had the lowest. The number of visits dependent on the day of the week is a well-known phenomenon.^[4]

Despite the considerably higher visit load on Sundays, the admission rate was not different from the overall average. However, the admission rate on Saturdays (the lowest visit load) was higher than the overall admission rate (30% vs. 28%). This may be attributable either to a more severe case mix (milder cases remained at home), less experienced staff on weekends) or less restriction to admit due to better availability. We evaluated the association between daily ED visits (independent of the day of the week) and the inpatient admission rate. The results revealed a mild dependency between ED visits and admission rate (r² = −0.14); hence, we may conclude that a lower visit load may enable higher bed supply and, accordingly, less restriction on inpatient admission.

The daily average duration to discharge was highest on Sunday and lowest on Saturday, which resembles the number of visits. We evaluated the association between the daily average duration and number of daily visits (regardless of the day of the week), which revealed a low association. This suggests that the underlying explanations are other than visit load per se.

Dropout rates showed neither dependency on the day of the week nor visit load; however, the daily dropout rate was associated with the daily duration to discharge. Duration to discharge may be considered a proxy for perceived inefficient processes, while dropout may be considered a proxy for dissatisfaction and intolerability.

6. Limitations

The study had based on administrative data hence it lacks clinical insights. The data is of a single tertiary medical center hence may represent local behavior and patterns that may not be generalized countrywide or global. The Israeli working week is different than most western countries as the weekend day off is Saturday, not Sunday. That also means that Sunday in Israel is the first working day after the weekend rather than Monday in western countries. The importance of the study is in its simple method that enables each hospital ED to appropriately prepare for expected visiting load and admission rates pattern that enables better allocation of staff and resources.

7. Conclusions

The only reliable predictor of ED visits load was the day of the week. It can be managed by appropriate staffing, allocation of more competent staff, and availability of beds. It may also reduce the duration of discharge and maintain better accessibility, timely care, patient satisfaction, and may accordingly decrease the dropout rate.

The relatively stable admission rates suggested well-established guidelines independent of visits number.

The high discharge rate (nearly 70%) suggests that community or ambulatory alternatives to ED may be sufficient to treat the majority of these patients. This may be achieved partially either through ED outpatient clinic or community services.

We recommended each medical center to evaluate its ED daily output measures dependency on seasonality and day of the week for operational optimization.

Author contributions

Uri Gabbay initiated and designed the study, performed data analysis, interpreted the findings, and drafted the manuscript.

Michael Dresher contributed to the study design, interpreted the findings, co-drafted the manuscript, and approved the final draft.