Examination of Impact of After-Hours Admissions on Hospital Resource Use, Patient Outcomes, and Costs

Background Nighttime and weekends in hospital and intensive care unit (ICU) contexts are thought to present a greater risk for adverse events than daytime admissions. Although some studies exist comparing admission time with patient outcomes, the results are contradictory. No studies currently exist comparing costs with the time of admission. We investigated the differences in-hospital mortality, ICU length of stay, ICU mortality, and cost between daytime and nighttime admissions. Methods All adult patients (≥18 years of age) admitted to a large academic medical-surgical ICU between 2011 and 2015 were included. Admission cohorts were defined as daytime (8:00–16:59) or nighttime (17:00–07:59). Student's t-tests and chi-squared tests were used to test for associations between days spent in the ICU, days on mechanical ventilation, comorbidities, diagnoses, and cohort membership. Regression analysis was used to test for associations between patient and hospitalization characteristics and in-hospital mortality and total ICU costs. Results The majority of admissions occurred during nighttime hours (69.5%) with no difference in the overall Elixhauser comorbidity score between groups (p=0.22). Overall ICU length of stay was 7.96 days for daytime admissions compared to 7.07 days (p=0.001) for patients admitted during nighttime hours. Overall mortality was significantly higher in daytime admissions (22.5% vs 20.6, p=0.012); however, ICU mortality was not different. The average MODS was 2.9 with those admitted during the daytime having a significantly higher MODS (3.0, p=0.046). Total ICU cost was significantly higher for daytime admissions (p=0.003). Adjusted ICU mortality was similar in both groups despite an increased rate of adverse events for nighttime admissions. Daytime admissions were associated with increased cost. There was no difference in all hospital total cost or all hospital direct cost between groups. These findings are likely due to the higher severity of illness in daytime admissions. Conclusion Daytime admissions were associated with a higher severity of illness, mortality rate, and ICU cost. To further account for the effect of staffing differences during off-hours, it may be beneficial to compare weekday and weeknight admission times with associated mortality rates.


Introduction
Te hours immediately following hospital and intensive care unit (ICU) admission are crucial for infuencing patient outcomes since vulnerability to complications is greatest at this time [1][2][3]. In this context, it is important that diagnosis and treatment plan formulation occur rapidly to reduce patient mortality, regardless of the time of admission [4,5]. Nights and weekends (of-hours) have been recognized as high-risk admission times for both hospital and ICU patients [4][5][6]. Typically, of-hours consist of fewer staf on-call, with a greater proportion of these staf being less experienced compared to daytime staf [7]. Few studies exist which examine the relationship between admission time and patient outcomes. Furthermore, existing study results are contradictory, perhaps because of diferences across study sites in terms of patient characteristics and stafng [7][8][9]. Studies conducted in Saudi Arabia, the United Kingdom, and Taiwan show that including more experienced intensivists and more detailed management guidelines during ofhours improved ICU patient outcomes on weekends and weeknights [4,10,11]. However, these studies tend not to account for diferences in illness severity or resources needed. Multiple studies have found no diference in mortality, with some studies showing a decrease in mortality rates for night time admissions [1,5,8,[11][12][13][14][15][16][17]. For studies analyzing admission times for single illnesses or injuries such as hip fractures or gastrointestinal haemorrhage, the results generally show signifcant diferences in patient outcomes between regular hours and of-hours, with patients admitted during of-hours having higher mortality rates [18,19].
We are not aware of studies comparing costs between daytime and nighttime admissions. However, it is an important question as it allows for further investigation into more cost-efective practice for daytime and nighttime ICU admissions.
Te objective of this study was to investigate the mortality rates between daytime and nighttime admissions as well as patient outcomes such as ICU length of stay, days on mechanical ventilation, the occurrence of adverse events, and cost.

Methods
Setting and patient collection: Tis retrospective cohort study included all adult patients (≥18 years of age) admitted to an ICU at Te Ottawa Hospital between January 1, 2011, and December 31, 2015. Patients were followed up for the length of their hospitalization until discharge or up to one year, whichever came frst, including transfers out of and back into the ICU if applicable. Patients were classifed according to the frst initial ICU admission time as daytime (08:00-16:59) or nighttime (17:00-07:59) admission cohorts, correlating with the presence or lack of an attending presence, respectively.
Data Collection: All data were obtained from the Ottawa Hospital Data Warehouse, a health administrative database that integrates several information resources including a clinical data repository and case-costing system [20,21]. Baseline patient demographic data, previous admission, transfer data, and patient diagnosis at the time of ICU admission were assessed in parallel for the daytime and nighttime patient cohorts. Comorbid conditions were summarized and presented as the Elixhauser (ELIX) comorbidity index that uses coded information based on the International Classifcation of Diseases, Version 10 (ICD-10-CA) [22,23]. Outcome data were collected from the time of admission until either the point of discharge from the hospital or death and included ICU length of stay, discharge data, and patient safety indicators, among others. Further, hospital costs accrued by patients including total hospital costs, direct costs, and costs of an ICU stay were assessed [21]. Total hospital costs included both direct and indirect costs. Direct costs refer to all expenses paid to the hospital with fee codes linked to the patient chart. Tis includes salaries and benefts for the management staf, equipment, screening, and materials but does not include physician's remuneration due to the billing structure within the province. Indirect costs refer to any overhead operational fees including the cost of the room occupied by the patient on an hourly basis. Case costing at the Ottawa Hospital is determined based on the Ontario case costing initiative [24] described in the Canadian Institute for Health Information Management Information Systems Guidelines [25]. All costs were reported in Canadian dollars (CDN).
Data Analysis: Statistical analyses were performed using GraphPad Prism Software v. 8.3.0 (538) and Stata v. 11. Continuous variables including age, ELIX score, days spent in ICU, and days on mechanical ventilation were compared between daytime and nighttime admission cohorts using the Student's t-test (parametric values) with a level of signifcance of a p value of equal to or less than 0.05. Tis data was presented as a calculated mean value followed by the standard deviation (SD). Patient data pertaining to cost analysis, including total ICU costs and total hospitalassociated costs (including direct costs) were assessed using a Student's t-test and presented as mean values followed by standard deviation (SD) values. Chi-square tests (χ 2 ) were used for the remaining variables to determine the diferences among patients between the daytime and nighttime patient cohorts. p value equal or less than 0.05 was considered statistically signifcant. Regression analysis was used to test for associations between patient age, gender, the Elixhauser Comorbidity Score, the Multiorgan Dysfunction Score (MODS), use of dialysis, mechanical ventilation, and the most responsible diagnosis (MRD) and (1) in-hospital mortality and (2) total ICU costs. MRD was captured by the ICD-10 code. Missing data for ICU interventions accounted for 18.6% of patients. Te analysis that included these variables was done with complete data only.
Tis study was approved by the Ottawa Health Science Network Research Ethics Board.

Results
During the study period, 14,265 patients were admitted to the ICU. Patient characteristics are summarized in Table 1. Te average age was 63.74, and 56% of patients were male, with no statistical diference between daytime versus nighttime admissions.
In the multivariable analysis, greater age, female gender, higher Elixhauser comorbidity scores, certain most responsible diagnoses (digestive, infectious, and respiratory), higher MODS, and mechanical ventilation were associated with higher odds of in-hospital mortality ( Table 3). In a similar multivariable analysis, higher Elixhauser comorbidity scores, certain most responsible diagnoses (infectious and respiratory), dialysis, and mechanical ventilation were associated with higher total ICU costs ( Table 3). Greater age was associated with lower total ICU costs in this multivariable analysis. Time of admission was not found to be associated with in-hospital mortality (Table 3).
Te total ICU cost was signifcantly higher in those admitted during the daytime (p � 0.0027) ( Table 4). However, there was no diference in all hospital total cost or all hospital direct cost between groups. Te adjusted total ICU cost for time of admission was similar for both groups (Table 3).

Discussion
Te frst hours in the ICU have a large infuence on patient outcomes [1][2][3]. Previous studies have noted that of-hour admissions carry higher risks for patients [4][5][6]. To our knowledge, no previous studies have compared costs between daytime and nighttime admissions. In this singlecenter, retrospective cohort study of a mixed ICU, we found the majority of admissions occurred during nighttime hours. A number of studies have reported more admissions occurring during nighttime hours [8,10,11,14,15]. Tis, however, varies, with some studies showing higher admissions during the day [1,6,12,26]. Overall, the Elixhauser score was similar between groups. Previous studies have shown that patients admitted during of-hours including nighttime and weekends had higher severity of illness scores; however, scores used difered between studies [13,16,[26][27][28]. Luyt 2007 found that patients admitted during the day had a higher severity of illness. Tis is in agreement with our study which demonstrated that those admitted during the daytime had a higher severity of illness as evidenced by their higher MODS. Tis is also in keeping with the overall crude mortality being signifcantly higher for those admitted during the day, compared to those admitted during nighttime hours (p � 0.012). Conficting results have been reported for mortality associated with time of admission. Te majority of studies demonstrated no diference between daytime and nighttime admission mortality rates [1,5,8,[11][12][13][14]. Two meta-analyses [4,7] supported this. Laupland 2008 showed a decrease in mortality for daytime admissions, and other studies showed increased mortality with after-hour admissions [8,26,28,29].
Tere was no diference in odds of ICU mortality or cost between groups. Tis is reassuring, as the majority of academic ICUs including ours do not have a board-certifed intensivist in house during nighttime hours, only residents and/or fellows. In other studies, no diference in mortality rate was found for of-hour admissions with 24/7 staf onsite [10] or with a staf intensivist on call [17,30,31].
Numerous factors likely play into this. Firstly, a boardcertifed intensivist is available on call 24/7, and patients admitted during of-hours are usually reviewed in person with a fellow or staf member within 8 hours. Sicker patients also tended to be admitted during the daytime, which allows for staf intensivist input. Te nursing ratio is the same during day and night and services such as endoscopy, consultants, and other procedures are available during ofhours. During the day, numerous obligations require time from ICU staf and residents such as teaching, discharging, procedures, family updates, and team rounds that are not generally required at night. Typically, more testing and invasive procedures are ordered during the daytime, which are associated with their own morbidity and mortality [15].
Te main concern in assessing the diference between day and nighttime admissions is that patients admitted at night with fewer staf and no intensivist in-house may be at an increased risk of adverse outcomes. Tis study demonstrates that nighttime admissions are not associated with  [33]. Studies have also demonstrated that even with only trainees in the ICU overnight, patient outcomes are not improved, and the number of adverse events does not decrease by having more physicians or more senior physicians in the ICU during the night [33][34][35].
While the higher severity of illness as evidenced by higher MODS, days of mechanical ventilation, requirement for dialysis, and length of stay is likely contributing to the higher in-hospital mortality rate for daytime admissions, other factors may also be contributing. For example, daytime allows for admission from elective surgery, during which complications may arise, and a signifcantly higher rate of transfers from the surgical ward occurred during the day in our study. Secondly, a higher number of transfers from other institutions occurred during the day and those who were admitted during the day were also more likely to require transfer to other acute care institutions or continuing care, pointing to the overall severity of illness. It may also be that due to logistical factors that admissions to the ICU during the day are delayed, and delayed admissions have been shown to increase mortality rates [4,36].
While previous studies have assessed the diference in mortality and patient outcomes between daytime and nighttime admissions, none, to our knowledge, have addressed possible cost diferences; however, with rising costs of healthcare, this is an important parameter to assess and may have potential practice-changing implications. Our study found that those admitted during the daytime had a signifcantly higher total ICU cost. Te adjusted ICU cost was similar between groups. Tis is likely due to the lower severity of illness for nighttime admissions. Tere was no diference between nighttime and daytime admission for overall hospital cost. Tis is likely explained by the increased severity of illness and mortality rates in those admitted Table 3: Multiple regression analysis between time of admission and mortality and total ICU costs.

Factors
In-hospital mortality Total ICU costs OR (95% CI) p value β coefcient (95% CI) p value during the daytime, therefore not requiring a prolonged hospital stay. Tis study has limitations including that it is a retrospective single-institution study along with the use of multiple comparisons. It did not compare medical versus surgical patients, as we have mixed ICUs. Secondly, intensivists were present during the day on weekdays and weekends, so it may have been benefcial to assess the mortality rate comparing weekdays vs weekends and/or weekday nights vs weekend nights to further assess the impact of the presence of a qualifed intensivist. Physician remuneration was not included in this study which may have altered the cost analysis.

Conclusion
Overall, we found that nighttime ICU admission is not associated with worse outcomes or higher mortality rates. Protective factors include the availability of procedures and consultants at all times, ICU staf on call 24/7, and the use of early ICU assessment through our ICU outreach rapid response team.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that there are no conficts of interest.