BACKGROUND: Critically ill patients are admitted to the pediatric ICU at all times, while staffing and other factors may vary by day of the week or time of day. The purpose of this study was to evaluate whether admission during off-hours is independently associated with mortality in PICUs.
METHODS: A retrospective cohort study of admissions of patients <18 years of age to PICUs was performed using the Virtual PICU Systems (VPS, LLC) database. “Off-hours” was defined as nighttime (7:00 pm to 6:59 am) or weekend (Saturday or Sunday any time). Mixed-effects multivariable regression was performed by using Pediatric Index of Mortality 2 (PIM2) to adjust for severity of illness. Primary outcome was death in the pediatric ICU.
RESULTS: Data from 234 192 admissions to 99 PICUs from January 2009 to September 2012 were included. When compared with regular weekday admissions, off-hours admissions were less likely to be elective, had a higher risk for mortality by PIM2, and had a higher observed ICU mortality (off-hours 2.7% vs weekdays 2.2%; P < .001). Multivariable regression revealed that, after adjustment for other significant factors, off-hours admission was associated with lower odds of mortality (odds ratio, 0.91; 95% confidence interval, 0.85–0.97; P = .004). Post hoc multivariable analysis revealed that admission during the morning period 6:00 am to 10:59 am was independently associated with death (odds ratio, 1.27; 95% confidence interval, 1.16–1.39; P < .0001).
CONCLUSIONS: Off-hours admission does not independently increase odds of death in the PICU. Admission from 6:00 am to 10:59 am is associated with increased risk for death and warrants further investigation in the PICU population.
- CI —
- confidence interval
- ED —
- emergency department
- OR —
- odds ratio
- PIM2 —
- Pediatric Index of Mortality 2
- PRISM III —
- Pediatric Risk of Mortality III
- VPS —
- Virtual PICU Systems database
What’s Known on This Subject:
Admissions to the ICU during off-hours (nights and weekends) have been variably associated with increased mortality in both adults and children. Changes in staffing patterns, patient characteristics, or other factors may have influenced this relationship over time.
What This Study Adds:
This study demonstrates in a large, current, multicenter database sample that off-hours admissions to PICUs are not associated with increased risk-adjusted mortality. Admissions in the morning from 6:00 am to 10:59 am are associated with increased mortality and warrant further attention.
Critically ill patients may be admitted to the ICU at any time, but care provided may vary by day of the week or time of day. Decreased staff numbers and experience level, as well as human factors such as fatigue1 and circadian rhythm disruption2 have been hypothesized to adversely affect outcomes of patients admitted during off-hours.3 A meta-analysis and a large national registry study of adult ICU admissions, each evaluating >130 000 admissions, demonstrated increased odds of death associated with weekend admission.4,5 In the largest study in the pediatric ICU setting to date, which included >20 000 admissions to 15 PICUs from 1995 to 2001, nighttime admission was associated with increased odds of death.6 Several smaller studies in PICUs found no difference in risk-adjusted mortality for patients admitted during off-hours when compared with weekdays,7–9 and 1 study found lower risk-adjusted mortality for admissions during off-hours.10
Reasons for the varied results in studies evaluating off-hours admission and mortality in the pediatric ICU population may include type II error related to smaller numbers of admissions and lower mortality rates compared with adult studies, heterogeneous samples, or changes in care over time (such as increased prevalence of 24-hour in-house attending intensivist coverage). Our objective in this study was to investigate the possible association between off-hours admission and mortality in a large, current sample of pediatric ICU patients. We hypothesized that mortality would be higher for off-hours admissions when compared with regular weekday admissions, even when adjusted for other significant factors. Additionally, we hypothesized that risk-adjusted mortality would be higher for the subsets of (1) weekend versus weekday and (2) night versus daytime admissions.
De-identified admission data were obtained from the Virtual PICU Systems (VPS, LLC) database for all admissions <18 years of age to a participating pediatric ICU in the United States from January 1, 2009 through September 29, 2012. This represented the most recent period for which data were available in the fields of interest at the time of data retrieval. The VPS database is a multicenter database of PICUs, defined as a physical space dedicated exclusively to the care of critically ill children, and is used primarily for benchmarking and quality assessments.11 VPS data were provided by the VPS, LLC. No endorsement or editorial restriction of the interpretation of these data or opinions of the authors has been implied or stated.
Data collected from the VPS database included academic center designation, 24-hour every day in-house (24/7) attending intensivist presence, demographics, admission diagnosis, origin of admission, trauma status, day and time of admission, severity of illness scoring, and ICU mortality. All variables were specific to each admission (eg, 24-hour, in-house intensivist designation for that center at the time of admission). The 8 most frequent primary diagnostic categories were retained for descriptive statistics (respiratory, cardiovascular, neurologic, injury/poisoning/adverse effects, hematology/oncology, orthopedic, infectious, and gastrointestinal), with the remainder being consolidated into 1 category, “other” (see Appendix). Admissions designated as post-cardiac bypass were excluded owing to concern for inability to adequately control for severity of illness.12 Origin of admission was grouped as emergency department (ED), inpatient (including ward and stepdown), operating room, post-anesthesia care unit, another ICU from the same hospital (neonatal ICU, delivery room, or adult ICU), outside hospital ICU, outside hospital ED, and “other” (see Appendix).
This was a retrospective cohort study. The primary outcome was ICU mortality. The exposure was off-hours admission, defined as admission occurring during nighttime (between 7:00 pm and 6:59 am the next day, regardless of day of week) or weekends (12:00 am Saturday until 11:59 pm Sunday). Regular weekdays were defined as Monday through Friday 7:00 am to 6:59 pm for comparison with off-hours. For secondary analyses the following comparisons were made: (1) admissions on weekends (12:00 am Saturday until 11:59 pm Sunday) were compared with weekdays (12:00 am Monday until 11:59 pm Friday) and (2) admissions during nighttime (7:00 pm until 6:59 am every day) were compared with daytime (7:00 am until 6:59 pm every day).
Severity of illness adjustment was performed by using Pediatric Index of Mortality 2 (PIM2),13 as this is a required data field in the database. PIM2 risk of mortality values were multiplied by 100 (converted to a percent value) for the purposes of scale. Pediatric Risk of Mortality III (PRISM III)14 data were also obtained when available and the primary analysis was again performed by using this adjustment for comparison. Evaluation of high-risk subgroups previously identified to have increased risk for death with nighttime admission6 included patients who had cardiac arrest before admission identified from the PIM2 high-risk diagnosis field, patients who had congenital cardiovascular disease identified from the VPS primary diagnosis subcategory field (excluding those immediately post-cardiac bypass as noted previously), and patients who had shock identified from ICD-9-CM codes (see Appendix).
VPS data are entered by trained individuals at each participating site, and quarterly reviews are performed to ensure data quality and inter-rater reliability. All variables in the primary analysis were required for VPS participants during the study period, thus there were 0% missing data for these fields. PRISM III was not a required field in the VPS database, however 91% of admissions had PRISM III data available.
Continuous descriptive data were compared by using Student’s t test for parametric data or Mann-Whitney U test for non-parametric data, where appropriate. Categorical data were compared using Pearson’s χ2 test.
Factors selected a priori to be evaluated in the analysis included off-hours admission, PIM2 risk of mortality, age category, gender, origin of admission,15 trauma status, and 24/7 in-house intensivist presence. Components of each severity of illness scoring tool were not used individually in the regression model, because the PIM2 risk of mortality entered in the model had been calculated using these elements. Primary diagnosis categories were also not used individually in the model because PIM2 risk of mortality entered in the model includes fields for high- and low-risk diagnoses. Mixed effects multivariable logistic regression with clustering at the hospital level was performed using all variables. Evaluation of condition index and variance inflation factor was performed to exclude collinearity. For secondary outcomes, a similar mixed-effects multivariable logistic regression was performed substituting nighttime or weekend admission, respectively, for off-hours admission. An additional analysis was performed using PRISM III score to adjust for severity of illness instead of PIM2 score for those centers reporting PRISM III data. Observed versus expected graphical analysis was performed by calculating actual mortality for patients admitted during each day of the week or hour of the day, and comparing with expected mortality by mean PIM2 score for all patients admitted during that day or hour, as appropriate. Graphical review of admissions by day of week and time of day was performed for each center to ensure that all units admitted patients during nights and weekends. Post hoc analysis was performed using admission during the hours of 6:00 am to 10:59 am on all days of the week as an independent variable (versus all other times of admission).
Data were analyzed by using SAS software version 9.2 (SAS Institute Inc, Cary, NC). The Institutional Review Board of Wake Forest University Baptist Hospital approved this study.
Retrospective data meeting inclusion criteria from 243 292 admissions to 99 centers were obtained from the VPS, LLC database. One admission was excluded owing to a negative entry for age, and 9099 were excluded owing to being designated as post-cardiac bypass, leaving 234 192 admissions in the final analysis. The median number of admissions per center was 2093 (interquartile range, 821–3486; range, 67–7984), and 66% were classified as academic centers. Centers ranged in capacity from categories of 0–7 to ≥30 beds; median category was 18 to 20 beds, and 70% of centers were designated as having 15 beds or more. In-house, 24/7 staffing by a pediatric intensive care attending was present in 55/99 (56%) centers for at least part of the study period (45% for the entire period).
Patient ages ranged from newborn through 17 years. Patient characteristics are summarized in Table 1. Overall ICU mortality was 2.4% for the sample. Mortality by day of admission is shown in Fig 1; weekend admissions had a higher observed mortality (3.1%) compared with weekday admissions (2.3%). Mortality by hour of admission is shown in Fig 2; nighttime admissions had a higher observed mortality (2.5%) than daytime admissions (2.4%).
Primary admission diagnoses are summarized in Table 2. For all time periods, respiratory causes were the most common reason for admission. Admissions classified as trauma were predominantly in the Injury/Poisoning/Adverse Effects category for primary diagnosis (81%), with some in the Neurologic category (11%), and the remainder divided across other categories.
Associations with ICU mortality on multivariable analysis are displayed in Table 3. Use of PRISM III rather than PIM2 risk of mortality in the multivariable model yielded a similar odds ratio estimate for off-hours admission (odds ratio [OR], 0.90; 95% confidence interval [CI], 0.83–0.97; P = .005). Secondary analysis of weekend (Saturday 12:00 am to Sunday 11:59 pm) versus weekday (Monday 12:00 am to Friday 11:59 pm) admissions yielded no significant difference in odds of mortality (OR, 1.01; 95% CI, 0.94–1.09; P = .79). Nighttime admissions (7:00 pm until 6:59 am every day) were associated with significantly lower odds of mortality versus daytime (7:00 am until 06:59 pm every day) (OR, 0.86; 95% CI, 0.81–0.92; P < .0001). Off-hours admission was not significantly associated with mortality when the multivariable model was applied to the subgroups of patients admitted with cardiac arrest (n = 4420; OR for off-hours admission 1.05; 95% CI, 0.89–1.23; P = .59), congential cardiovascular disease (n = 5962; OR, 1.04; 95% CI, 0.73–1.49; P = .81), or shock (n = 3475; OR, 0.78; 95% CI, 0.59–1.03; P = .08).
A post hoc analysis was performed to evaluate the morning hours of admission between 6:00 am and10:59 am, which were associated with peak observed mortality (unadjusted mortality 3.2% versus all other times 2.3%). Admissions during this morning period also had the highest expected mortality by PIM2 (Fig 2). On weekends there was a more extended period from 7:00 am to 2:59 pm during which admissions were associated with higher mortality (Fig 2). Admissions during the 7:00 am hour were associated with the highest unadjusted ICU mortality on weekdays (3.5%), whereas mortality was highest for admissions during the 9:00 am hour on weekends (4.4%). A multivariable regression analysis across the entire sample for admission during the period of 6:00 am to 10:59 am, compared with all other hours, revealed an OR for ICU mortality of 1.27 (95% CI, 1.26–1.39; P < .0001).
A review of admission trends revealed that the morning period of 6:00 am to 10:59 am accounted for only 24 715 (10.6%) admissions (if all hours had equal admission density, a 5-hour period would be expected to contain ∼20% of admissions), with the lowest percentage of admissions during the day occurring during the 07:00 am hour (Fig 3). During the period from 6:00 am to 10:59 am, admissions from the inpatient floor accounted for a significantly larger proportion of total admissions (morning, 24.4% of admissions versus all other times: 12.9%; P < .0001) and deaths (morning, 27.2% of patients who died were admitted from inpatient floor versus all other times: 16.7% of patients who died were admitted from inpatient floor; P < .0001). A post hoc evaluation of morning admissions who died versus survived revealed no significant differences associated with minimum nighttime ICU coverage models (nighttime attending intensivist model in place for 58% of morning admissions who died versus 58% who survived, P = .64; fellow, 25% vs 24%, P = .43; resident, 9.3% vs 10.5%, P = .30; hospitalist, 0.6% vs 1.2%, P = .18; attending from home only, 6.5% vs 6.8%, P = .78).
This study reports on the largest cohort to date on this topic and demonstrates that admission to the pediatric ICU during off-hours is independently associated with decreased risk of mortality. These findings are contrary to the original hypothesis for the study, and discordant with the increased risk for death for nighttime admissions found in the largest previous multicenter study in the pediatric ICU population.6 Our study demonstrated a decreased risk for death specifically for nighttime admissions compared with daytime, whereas weekend admissions were no different from weekdays on multivariable analysis. This finding is similar to a report from a single pediatric ICU in Australia, in which admissions during off-hours had an OR for mortality of 0.71 (95% CI, 0.52–0.98).10
Our finding of decreased odds of death for nighttime admissions may simply reflect that the highest risk period for our cohort was actually 6:00 am to 10:59 am, resulting in the majority of the higher-risk period being in the daytime cohort. Only 10.7% of admissions occurred during this 5-hour morning period, demonstrating that high volume of admissions did not contribute to the finding, but raising the concern that the results may be skewed owing to the relatively small sample during those hours.
Alternatively, there may be variations in care contributing to decreased risk for death for nighttime admissions and increased risk for morning admissions. The increased proportion of transfers from the inpatient floor during the morning period of 6:00 am to 10:59 (and deaths among these patients) raises the concern for clinical deterioration overnight that was under-recognized, leading to delayed transfer to the ICU until the morning and increased mortality. Nighttime caregivers on the inpatient floors (such as nurses and resident physicians) may be less experienced and have increased workload, leading to delayed recognition of need for transfer to the ICU. This could be particularly problematic in patients having progression of their original disease process and/or responding poorly to clinical interventions during their inpatient stay.15 The extent of illness of such patients may not be completely captured using PIM2 or other scoring systems. Furthermore, the later hour associated with peak mortality for weekend admissions (peak mortality for admissions at 9:00 am on weekends versus 7:00 am on weekdays) could be attributable to later arrival of daytime caregivers to assess patients on the inpatient floors. A similar phenomenon could also delay patient presentation to the ER overnight (with subsequent delay of ICU transfer), as caregivers may be sleeping or otherwise less disposed to bring their child to medical attention late at night. Information about nighttime hospital floor coverage, rapid response team activation, or ER admission patterns was not available for our sample, but may warrant further evaluation in future studies on this topic.
Variations in care within the ICU, such as transitions of care and dilution of caregiver attention, could also contribute to the decreased odds of death for nighttime admissions with increased risk for morning admissions. Hand-offs between physicians, nurses, and other care providers often occur during the 7:00 am hour, which was associated with peak mortality (3.8%) for the overall sample. However, no such increase in mortality was observed around the 7:00 pm hour when similar transitions may occur on weekdays; there was a small increase at 7:00 pm noted on weekends. At night, there are fewer admissions overall and reduced activities such as rounds, conferences, or transports out of the unit for diagnostic or therapeutic procedures, which could compete for the attention of daytime caregivers. The period from 6:00 am to 10:59 am includes a time period when morning rounds are conducted in most ICUs. In a study of medical and surgical admissions to 4 adult ICUs, Afessa et al16 reported that admission during the rounding time from 8:00 am to 10:59 am was independently associated with increased risk for hospital death, with an OR of 1.32 (95% CI, 1.18–1.48). Similarly, de Souza et al17 evaluated medical and surgical admissions to 5 adult ICUs and found that admission between 8:00 am and 11:59 am was associated with increased severity of illness and increased odds of ICU mortality (OR, 1.19; 95% CI, 1.03–1.38). Several other studies have shown peak mortality for adult ICU patients admitted in the morning,5,18,19 however, no studies in the pediatric ICU population to date have reported a similar finding. The observed difference in time associated with peak mortality on weekends (9:00 am) versus weekdays (7:00 am) could relate to later rounding time in the ICU on weekends. Future studies are warranted to determine whether an association with increased mortality with morning admissions persists across other samples and what factors may be contributory.
Factors independently associated with increased risk for mortality in our study included neonatal and infant age groups, certain origins of admission (particularly inpatient locations or other ICUs), and being a trauma admission. Age has not uniformly been associated with mortality previously,6,13 however, neonatal and infant age groups appear to often be at higher risk when age is stratified into these groups.9,20,21 Increased risk for death for ICU admissions from inpatient locations has been previously described,6,15 and may represent additional patient decompensation despite interventions over time, which is not fully captured by severity of illness scoring.
Mortality is relatively rare in the pediatric ICU and is influenced by a myriad of factors that may or may not be related to factors at the time of admission. Among those who died, median length of stay until death was 3.9 days for regular weekday admissions and 3.1 days for off-hours admissions. It is unclear over what time period any variation in care at the time of admission is likely to influence mortality, and likely varies widely among patients and diagnoses. Our study did not evaluate other clinically important outcomes other than mortality, or aspects of off-hours care at later times during the ICU stay. Cardiac arrests occurring in the ICU during nights and weekends have been associated with decreased survival in both adults and children.22,23 Medication errors24 and complications during intubation25 in pediatric inpatients have also been reported to be more common during off-hours. Although 24/7 in-house intensivist coverage at the time of off-hours admission did not affect mortality in our study, transition to 24/7 in-house intensivist staffing has been associated with decreased duration of mechanical ventilation and ICU length of stay at a large tertiary care PICU,26 implying that improved care during off-hours could affect ventilator weaning and other factors over the course of a patient’s stay.
A major strength of our study is the large number of admissions included across 99 PICUs, including both academic and non-academic centers and many centers with 24-hour, in-house intensivist staffing. The PICUs in the United States included in this study represent 29% of the ∼337 PICUs in the United States.27 Another strength of our study is the data quality; only 1 admission was excluded owing to an erroneous value for age, and complete data were available for every independent variable included in the primary analysis.
Limitations of our study are primarily related to the confines of using a large multicenter dataset. We did not have information about deaths after ICU discharge (such as on the hospital ward, at home, or in hospice care). Information about limitations in care at the time of admission were not available. Protocols that may have been present at certain centers such as caps on nighttime admissions, separate admitting teams, or extra physician staffing for admissions during rounds or other times, were not available. Details of nurse and advance practice provider staffing were not available. Increased nurse staffing has been associated with improved outcomes in the ICU setting, including reduced mortality,28,29 so it is possible that an association between off-hours admissions and mortality could exist in a subset of centers with decreased nurse staffing during these time periods.
Our study provides evidence in a large, current sample of PICUs that off-hours admission is independently associated with decreased mortality. Future studies are warranted to evaluate admissions during the morning time period of 6:00 am to 10:59 am, which were associated with peak mortality in our study, as well as other clinically important outcomes that could be associated with variations in care during off-hours.
Appendix Data Groupings
Primary Diagnosis Category
“Other” category included those categorized as: Dermatologic, Endocrinologic, Factors Influencing Health, Genetic, Gynecologic, Immunologic, Metabolic, Newborn/Perinatal, Ophthalmologic, Psychiatric, Renal/Genitourinary, Rheumatologic, Symptoms, Transplant, and Ungroupable.
Origin of Admission
“Other” category included those categorized as: outside hospital inpatient, outside hospital stepdown, outside hospital operating room, outside hospital clinic, home, inpatient or outpatient procedure suite, chronic care facility, physical or pulmonary rehabilitation facility, psychiatric or substance abuse facility, and other.
“Shock” included ICD-9-CM codes specifying: “Anaphylactic shock,” “Anaphylactic shock-serum,” “Cardiogenic shock,” “Postoperative shock,” “Septic shock,” “Shock due to anesthesia,” “Shock nos,” “Shock w/o trauma NEC,” “Toxic shock syndrome,” and “Traumatic shock.” This included all ICD-9-CM codes in the data set with the word “shock” in the title; other codes that may have indicated shock (eg, Gram-negative septicemia) were excluded for specificity, as the purpose was to evaluate the most critical patients in shock.
- Accepted August 7, 2014.
- Address correspondence to Michael C. McCrory, MD, MS, Department of Anesthesiology, Section on Pediatric Critical Care Medicine, Wake Forest School of Medicine, Medical Center Blvd, Winston-Salem, NC 27157-1009. E-mail:
Dr McCrory designed the study, obtained the data, performed the statistical analyses, drafted the initial manuscript, prepared the tables and figures, and revised the manuscript; Drs Gower and Simpson assisted with the statistical design and analyses (including regular meetings to discuss the statistical approach used, assisting with data cleaning and data screening, providing and reviewing code for the statistical software program, reviewing output from the statistical software program) and reviewed and revised the manuscript; Drs Nakagawa, Mou, and Morris assisted with the research design and plan for data retrieval and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Slonim AD,
- Khandelwal S,
- He J,
- et al
- Miller AD,
- Piro CC,
- Rudisill CN,
- Bookstaver PB,
- Bair JD,
- Bennett CL
- Copyright © 2014 by the American Academy of Pediatrics