Factors and outcomes associated with paediatric emergency department arrival patterns through the day

Abstract

Introduction

Steadily increasing emergency department (ED) utilization has prompted efforts to increase resource allocation to meet demand. Little is known about the distribution and characteristics of patient arrivals by time of day. This study describes the variability and patterns of ED resource utilization related to patient, acuity, clinical, and disposition characteristics over a 24-hour period.

Methods

Retrospective cross-sectional study of all visits to a tertiary children’s hospital over a 1-year period. We use descriptive statistics to present ED visit details stratified by shift of arrival, and multivariable regression to explore the association between shift of presentation and hospital admission at index and 7-day return ED visits.

Results

Of 46,942 visits during the study period, 12% arrived overnight, 42% during the day, and 45% during the evening with variability in pattern of shift arrival by day of week. Overnight arrivals had a higher acuity (Canadian Triage and Acuity Scale [CTAS]) and different presenting complaints (more viral infection, less minor trauma) than day and evening arrivals, but similar ED length of stay. Shift of arrival was not associated with admission to hospital, but age, gender, socioeconomic status (SES), and day of week were.

Discussion

ED utilization patterns vary by shift of arrival. Though overnight arrivals represent a smaller proportion of total daily arrivals, their acuity is higher, and the spectrum of disease differs from day or evening arrivals.

Conclusions

Understanding variations and patterns of ED utilization by shift of arrival and day of week may be helpful in tailoring resource allocation to more accurately and specifically meet demands.

BACKGROUND

According to Canadian Institute for Health Information (CIHI) statistics, patients’ visits to emergency departments (EDs) have increased in the last few years (1). Similar patterns have been observed in other countries like the USA (2) and the UK (3). The rise in patient volume causes delays in care to both high and low acuity patients (4). The traditional response has been to increase the level of staffing in EDs (5,6) or introduce after-hour primary care services (7) to match volume demands. However, increasing staffing levels based solely on volume without consideration of acuity, arrival time, and variation over a 24-hour period may not optimize the impact of resource allocation.

There are only a few reports of ED describing the distribution of patient arrival by time of day. A study by Weir et al. examined the use of emergency services by children in southern Ontario (7) observing a gradual increase in frequency of visits between 8:00 and 23:00, with the highest frequency (34% of all visits) between 18:00 and 23:00 and only 16% of all visits presenting between 23:00 and 07:59. Downing et al. looked at time of ED visit among children in 13 different rural and urban EDs in the West Midlands region in England and found a similar pattern with the highest proportion of visits (15%) arriving between 18:00 and 20:00. They also observed fewer overnight visits with <12% of all visits occurring between midnight and 08:00 (8). Sacchetti et al. reported that paediatric visits to the ED increase during the evening hours (16:00 to 23:59) and account for 53% of total daily visits while only 10% presented between 24:00 and 07:59 (9).

While a handful of studies describe the basic epidemiology of ED arrivals by time of day, fewer report on the acuity or disposition from the ED. One retrospective study of children admitted to the intensive care unit (ICU) in London, Canada found that a quarter of the children admitted to the ICU had originally presented to the ED overnight, 41% had arrived between 08:00 and 16:00 and 33% between 16:00 and 24:00 (10). Unfortunately, neither the distribution of patient arrivals nor the relative acuity across shifts was described in this study, making conclusions relevant to staffing difficult to draw.

Despite evidence suggesting that paediatric ED patient volumes are proportionally reduced overnight, information about visit acuity, type of clinical presentation, and ED disposition to guide staffing models are still needed. While the ED must be staffed 24 hours a day to ensure patient safety and satisfaction (11), given the inherited disadvantages associated with overnight shift work such as fatigue (12,13), lack of resources (14), increase medication errors (15), and the shortages of ED physicians (16), simply increasing ED physician coverage across 24 hours might not be optimal solution.

The objective of this study is to describe the varying resource utilization of a paediatric ED over a 24-hour period, specifically related to acuity, range in the clinical conditions, disposition, and measures of flow. Identifying detailed ED utilization and needs throughout the day can guide resource allocation and help optimize staffing coverage, both in terms of numbers but also provider type, potentially allowing for improved use of nurse practitioners and physician assistants.

METHODS

Design, population, and setting

We conducted a retrospective cross-sectional study, analyzing administrative data from the BC Children’s Hospital (BCCH) paediatric ED. BCCH is the only tertiary care paediatric referral centre in British Columbia, and receives approximately 50,000 annual visits. The BCCH ED provides care to children and youth up to 17 years of age. During the study period, the BCCH ED was staffed with paediatric or general emergency physicians with single coverage overnight from 02:00 to 10:00, and double or triple coverage from 12:00 to 01:00 depending on the season (higher staffing in the winter and spring, lower in the summer and fall). Nursing coverage is constant across the year and generally includes 10 nurses on during the day, 10 during the evening, and 8 overnight. We included all patient visits to BCCH paediatric emergency department (PED) from September 1, 2015 to August 31, 2016 without exclusion.

Objectives and outcome measures

Our primary objective was to describe ED visits (volume and characteristics) stratified by shift of arrival: overnight (24:00 to 08:00), day (08:00 to 16:00), and evening (16:00 to 24:00). Visit characteristics included the acuity of visits categorized using the paediatric Canadian Triage and Acuity Scale (CTAS) score, patient sex, age, and reason for visit.

Our secondary objective was to explore the association between shift of presentation and (a) visit disposition outcomes (admitted versus routine discharge); (b) return to the ED resulting in an admission within 7 days of original ED discharge; and (c) visit length of stay (LOS), adjusting for acuity of visit and other co-variables such as age, gender, season of arrival, and socioeconomic status derived from the patient’s residence using postal code.

The study protocol was reviewed and approved by the BC Children’s and Women’s Research Ethics Board.

Data analysis

We summarized demographic data using descriptive statistics. Continuous variables were presented using means and standard deviations or medians and interquartile ranges when appropriate, and categorical variables were presented as proportions with 95% confidence intervals. Associations between the shift of presentation and disposition outcomes (admission versus discharge) or LOS were examined using multivariable regression analyses.

RESULTS

Study population and characteristics

There was a total of 46,942 visits to BC Children ED during the study period. Table 1 summarizes the patient and visit characteristics stratified by shift of presentation. Approximately 12% (5,618 patients) of the total volume presented during the night shifts (00:00 to 08:00). Day and evening arrival contributed to 42% and 45% of all visits, respectively. Patients arriving at night were slightly younger with a mean age of 4.9 years (standard deviation [SD] 5.1) than those coming during day and evening shifts with mean ages of 5.8 years (SD 5.2) and 5.5 (SD 5.1), respectively (P<0.001). Socioeconomic status did not appear to differ by shift of arrival and most patients were in the middle- to lower–middle-income range. Figure 1 depicts the acuity distribution by shift of arrival. Overnights had a relatively higher acuity of presentations, with 23% of visits triaged as CTAS 1 and 2 compared to 16% and 19% during day and evening shifts, respectively (P<0.001). We also observed a slightly larger proportion of visits resulting in hospitalization among night shift arrivals (P=0.0035) (Table 1). There was no meaningful difference in median ED LOS (interquartile range) between arrivals at night 2.5 hour (1.5 to 4.1), day 2.7 hours (1.7 to 4.2), or evening 2.8 hours (1.8 to 4.1). The most common reason for visits in the day and evening were minor trauma (e.g., lacerations and musculoskeletal injuries), whereas viral infections, including croup, were more common during the night shift (Table 2).

Table 1.

Demographic and visit characteristics by shift of arrival to the ED

	Night 0000–0800	Day 0800–1600	Evening 1600–2400
ED arrivals N (%)	5,618 (11.97%)	19,802 (42.18%)	21,523 (45.85%)
Mean Age in years (SD)	4.9 (5.1)	5.8 (5.2)	5.5 (5.1)
Sex (% Male)	56%	56%	55%
Triage N/%
CTAS 1	60 (1%)	95 (0.5%)	155 (0.7%)
CTAS 2	1,199 (21.5%)	3,004 (15%)	4,049 (19%)
CTAS 3	2,311 (41%)	7,399 (37%)	7,558 (35%)
CTAS 4	1,978 (35%)	8,810 (44.5)	9,387 (44%)
CTAS 5	40 (1%)	438 (2%)	307 (1.5%)
LOS median (IQR)	2.6 (1.6, 4.2)	2.8 (1.8, 4.3)	2.9 (1.9, 4.2)
Disposition
Hospitalized N (%)	510 (9%)	1,414 (7%)	1,650 (8%)
LWBS N (%)	265 (5%)	483 (2.5%)	1,252 (6%)
Total Annual Income N (%)
<40K	75 (1.3%)	230 (1%)	240 (1%)
40– <60K	1,992 (35.5%)	5,885 (30%)	6,947 (32%)
60– <90K	3,037(54%)	11,650 (59%)	12,292 (57%)
90– <125K	454 (8%)	1,771 (9%)	1,801 (8%)
>125K	10 (0.2%)	57 (0.3%)	34 (0.2%)

CTAS Canadian Triage and Acuity Scale; ED Emergency department; IQR Interquartile range; LWBS Left without being seen; SD Standard deviation.

Figure 1.

Emergency department triage acuity distribution per shift of arrival.

Table 2.

Ten most common clinical conditions associated with ED shift of arrival

Night 0000–0800 (N/%)	Day 0800–1600 (N/%)	Evening 1600–2400 (N/%)
1. Viral Infection (1,222/22%)	1. Minor Trauma (4,251/21%)	1. Minor Trauma (4,929/23%)
2. Gastrointestinal (588/10%)	2. Viral Infection (3,208/16%)	2. Viral Infection (3,384/16%)
3. Fever (450/8%)	3. Gastrointestinal (1,586/8%)	3. Gastrointestinal (1,754/8%)
4. Croup (436/8%)	4. Respiratory NOS (1,075/5%)	4. Fever (1,127/5%)
5. Minor Trauma (361/6%)	5. Skin rash (1,009/5%)	5. Skin rash (1,080/5%)
6. Gastroenteritis (307/5%)	6. Gastroenteritis (910/5%)	6. Respiratory NOS (1,054/5%)
7. Respiratory NOS (303/5%)	7. Fever RO sepsis (720/4%)	7. Gastroenteritis (871/4%)
8. Ear disorder (244/4%)	8. Neurology (710/4%)	8. Neurology (645/3%)
9. Skin Rash (211/4%)	9. Skin & soft tissue infections (644/3%)	9. Skin & soft tissue infections (574/3%)
10. Asthma (186/3%)	10. Ear disorder (509/3%)	10. Ear disorder (552/3%)

ED Emergency department; NOS Not otherwise specified.

Further evaluation of arrival patterns grouped by day of the week found variability in shift of arrival distribution between days of the week (P<0.001) as illustrated in Figure 2. While the proportion of visits arriving at night was consistent throughout the week, the difference between day and evening arrivals was more pronounced during weekdays than during the weekend, when day and evening arrivals were more evenly distributed. Overall, ED visits were highest on Sunday and Monday and lowest on Thursday and Friday. We did not observe any meaningful seasonal variations in the shift of arrival pattern.

Figure 2.

Shift of arrival frequency by day of week.

Secondary outcomes

We did not find any meaningful or statistically significant association between shift of arrival and hospitalization at the index after adjusting for covariables. We did find that higher acuity (CTAS 1, 2, and 3 as compared with CTAS 4 or 5), being female, increasing patient age and coming to the ED on a Friday appeared to increase the odds of hospitalization at index visit. We also observed lower odds of hospitalization during the winter, spring as compared to summer. Table 3 presents the adjusted odds ratios and 95% confidence intervals of admission to hospital at the index and within seven days of ED discharge by shift of arrival.

Table 3.

Association between shift of ED arrival and hospitalization at the index and 7-day return ED visits

	Adjusted OR (95% CI) Index ED visit	Adjusted OR (95% CI) 7-day return ED visit
Age	1.04 (1.04, 1.05)	1.00 (0.98, 1.01)
Sex
Male	Ref	Ref
Female	1.09 (1.01, 0.18)	1.05 (0.86–1.29)
SES
(<40k)	Ref	Ref
40–60K	0.84 (0.60, 1.17)	0.59 (0.30, 1.16)
60–90 K	1.1 (0.79,1.54)	0.53 (0.27, 1.04)
90–125 K	0.66 (0.46,0.95)	0.40 (0.19, 0.87)
>125 K	1.43 (0.71, 2.90)	0.40 (0.11, 1.49)
Triage acuity level
CTAS 4–5	Ref	Ref
CTAS 1	289.23 (216.11, 387.08)	6.96 (1.67, 28.95)
CTAS 2	55.147 (46.1, 65.97)	9.76 (7.14, 13.34)
CTAS 3	11.54 (9.62, 13.85)	4.39 (3.23, 5.95)
Shift of arrival
Night	Ref	Ref
Day	0.99 (0.88, 1.11)	1.04 (0.76, 1.44)
Evening	0.93 (0.83, 1.04)	1.08 (0.79, 1.48)
Seasons
Summer	Ref	Ref
Fall	0.90 (0.80, 1.00)	1.11 (0.81, 1.52)
Winter	0.75 (0.68,0.83)	1.13 (0.84, 1.52)
Spring	0.85 (0.76, 0.94)	1.20 (0.89, 1.63)
Day of week
Sunday	Ref	Ref
Monday	1.06 (0.92, 1.21)	0.85 (0.58, 1.26)
Tuesday	1.10 (0.96, 1.26)	0.97 (0.66, 1.41)
Wednesday	1.07 (0.93, 1.23)	1.04 (0.72, 1.52)
Thursday	1.02 (0.88, 1.17)	1.09 (0.75, 1.57)
Friday	1.17 (1.02, 1.34)	1.09 (0.75, 1.58)
Saturday	1.04 (0.89, 1.20)	1.19 (0.83, 1.71)

CI Confidence interval; CTAS Canadian Triage and Acuity Scale; ED Emergency department; OR Odds ratio; SES Socioeconomic status.

Among patients discharged at the index visit, we did not observe any association between shift of arrival and adjusted odds of hospitalization within seven days. The only covariable associated with increased odds of hospitalization within seven days of ED discharge was the acuity of the index ED visit.

DISCUSSION

In this study of resource utilization of a paediatric ED by shift of arrival, we found significant variation in utilization by time of day, with the lowest proportion of arrivals overnight and similar daytime and evening arrivals. We also noted variability in utilization by day of week with greatest utilization on Sunday and Monday and a higher proportion of evening arrivals during weekdays compared with weekends. Overnight arrivals had a higher acuity and slightly different epidemiology of presenting complaints compared with arrivals during the day and evening, but similar ED LOS and disposition outcomes after adjusting for covariables. While patient demographic (age, gender, and socioeconomic status [SES]), triage acuity (CTAS), and daily and seasonal factors were associated with hospitalization at the index or return ED visit within 7 days, we observed no differences in disposition based on shift of arrival. To our knowledge, this is the first study to report detailed analysis of patient, presenting and discharge data in relation to shift of patient arrival.

In our cohort, 12% of the patients arrived at the paediatric ED overnight, which is consistent with previous studies reporting a range from 10% to 16% (8–10). We found no significant variation in the average proportion of overnight arrivals by day of week or season of year, though the absolute patient volumes including overnight were highest in winter, consistent with a European paediatric emergency study by Massin et al. (17) in which the authors observed the night shift to be proportionally busier in winter than other seasons.

We were surprised to find that the average proportion of arrivals across the study period was almost equal between day and evening (42.2% versus 45.9%), as providers often express the perception that evenings are considerably busier. This perception may reflect an accumulation of patients waiting to be seen from daytime arrivals with relative understaffing during the daytime, however, we were unable to measure ED occupancy to test this hypothesis. When we compared the proportion of patient arrivals by shift and day of week, however, we found that the proportion of patients arriving in the evening was higher during the weekdays than weekend. This likely reflects parental work commitments or other barriers to daytime access during the week and may have implications with regard to staffing.

Despite the constant and relatively low proportion of patients arriving to the ED overnight, we found that the acuity and reasons for ED visit differed by shift of arrival, with a greater proportion of CTAS 1 and 2 patients and viral illness arriving overnight compared to day and evening, and more medical illness, and less minor trauma. No previous study has compared patient triage acuity or presenting complaint by time of patient arrival. These findings might influence the way in which ED staffing incorporates midlevel providers, matching them with these patterns based on scope of practice or role within the ED.

Only one prior study specifically has described disposition outcomes related to ED arrival time (11). In that study, the lowest proportion of paediatric ED admissions to ICU occurred during the overnight shift and the highest during the daytime. The authors of the previous study, however, did not provide details as to total arrivals by time of day, nor triage acuity, making direct comparison to our results and inferences related to ED resource utilization difficult. While our unadjusted data suggested a higher proportion of hospital admissions (not ICU specifically) during daytime hours, this association was not significant after adjusting for other variables. Despite observed differences in patient volume, acuity, and epidemiology, we found no difference in median ED LOS by shift of arrival.

Although shift of arrival was not related to disposition in our study, we did observe several factors associated with ED admission to hospital at the index visit. Nonmodifiable patient factors associated with increased odds of admission at the index visit included increasing age, male sex, while higher SES decreased the odds of admission. This last observation may reflect the social contributors to disease or different thresholds for ED utilization based on SES. We also noted that the highest odds of admission on index ED visit occurred during the summer months and this might reflect different epidemiology or hospital occupancy rates seasonally. Friday ED arrivals were associated with increased odds of admission, possibly as a result of the inability to assure timely follow-up over the weekend which might alter provider admission thresholds. Lastly, triage acuity had the highest association with hospital admission which further validates CTAS as a marker of acuity of illness.

Finally, similarly to previous studies on ED return visits, we found that admissions among those with a return ED visit within 7 days of discharge from the index visit were associated with nonmodifiable factors such as age, triage acuity, and SES (18). Though older age was predictive of index visit admission, younger age was associated with admission on return ED visit. SES had a similar effect on both index and return visit admissions. Interestingly, children triaged as CTAS 3 and 4 had an increased odd of admission on return visit within 7 days. This may reflect the greater diagnostic and disposition uncertainty among this cohort of patients, or the greater likelihood of admission at the index visit for those triaged as CTAS 1 and 2.

This study has several limitations. This was a single centre experience at a tertiary children’s hospital, and the results may not be generalizable to other areas or community hospitals with differing staffing and epidemiology. However, unpublished data suggest no significant differences in volume, triage acuity, admission rates, return visits, or mortality across eight paediatric EDs in four other provinces nor significant differences from the large community PED in the lower mainland of BC. Our observation that median LOS did not differ by shift of arrival did not take into account ED or hospital occupancy, or ED staffing data as these were not available and shift-based differences in ED flow may exist but were not detected using this single unadjusted measure. Finally, analyses of return ED visits were limited to those patients who returned to the study hospital and may have missed patients who returned to another facility.

CONCLUSION

We describe details of paediatric ED utilization by shift of arrival that may be useful in planning resource deployment strategy. While overnight arrival volumes are significantly lower than those during the day and evening, acuity and epidemiology varied by shift and may have implications when considering staffing models that include midlevel providers. Variations in patient arrival and disposition by day of week and season of year can further aid in the planning of staffing models that better meet the utilization demands.

Financial support: This research received no specific grant from any funding agency, commercial or not-for-profit sectors. QD is supported by a Michael Smith Foundation for Health Sciences Scholars salary award.

Potential Conflicts of Interest: All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.