Abstract
Background:
Recent interest in the ‘weekend effect’ has been expanded to cardiovascular intensive care units, yet the impact of off-hours admission on mortality and cardiovascular ICU (CICU) length of stay remains uncertain.
Objectives:
We examine the association between CICU admission day and time with mortality. Additionally, length-of-stay was also evaluated in relation to admission time.
Methods:
A single-center, retrospective cohort study was conducted including 10,638 adult patients admitted to a CICU in a tertiary-care academic medical center from July 1, 2012 to June 30, 2019. ICU mortality and length-of-stay were assessed by admission day and time adjusting for comorbid conditions and other clinical variables. We used logistic regression models to evaluate the factors associated with mortality and a generalized linear model (GLM) with log link function and gamma distribution was used to evaluate the factors associated with ICU length of stay.
Results:
Compared to weekday-day admissions, we observed an increased mortality for weekend-day for all admissions (6.5 vs 9.6%, Adjusted OR: 1.32 (1.03–1.72)), and for medical CICU admissions (7.6 vs 9.9%, Adjusted OR: 1.35 (1.02–1.79)). Additionally, compared to weekday-day, weekday-night admission was associated with 7% longer ICU length of stay in surgical ICU patients, 7% shorter length of stay in medical ICU patients.
Conclusion:
Admission to this open-model CICU during weekend hours (Saturday 08:00-Sunday 17:59) versus nights or weekdays is associated with increased mortality. ICU staffing care models should not significantly change based on the day of the week.
Keywords: Cardiovascular ICU, Mortality, Weekend admission, APACHE III
Introduction
Much interest surrounds hospital and intensive care unit (ICU) admission timing, demographics, and outcomes. In particular, differences in weekend mortality, the so-called “weekend effect,” has been a matter of special concern in the literature. A longstanding association between excess mortality and weekend versus a weekday hospital admission has been described in general hospital admissions. This phenomenon has also been explored in ICU patients.1–4 Additional factors including admission time, intensivist staffing, and open model ICUs, for example, have been implicated to affect mortality as well. Hospitals and health delivery systems have dealt with these issues in multiple ways, however policies directing changes in health care delivery were often difficult to adapt at a systems level to individual health systems and there remains significant variation in how ICU care is delivered.5 Admission timing has been expanded to study these ‘off-hours’, including both nights and weekends, where staffing and seniority were felt to play a role in the findings of excess mortality.6–9 Additionally, ICUs are often staffed 24 h a day by an intensivist, with data supporting improved outcomes in intensivist-staffed units.10,11 Much less data, however, exists in specific cardiovascular ICU (CICU) literature. These units represent a unique hybrid, often a combination of medicine and surgery patients in open or closed units with variable intensivist staffing. Recent efforts have attempted to define this population and derive novel risk predictors.12–17 This work defines the demographics of one CICU, then identifies and quantifies the effects of patient and clinical characteristics on mortality and CICU length of stay as a function of the admission day and time.
Methods
This retrospective cohort study was approved by ChristianaCare’s Institutional Review Board under waiver of informed consent as posing minimal risk to patients. The study was conducted in a 900-bed regional tertiary-care academic referral center in Newark, Delaware. Over 1500 patients are admitted yearly to our CICU. Nursing ratio averages 1:2–1:1 determined by acuity, independent of time of day or week. Respiratory therapists and pharmacists are available in-house 24 h daily. Allied health professionals including physical, occupational, and speech therapists; social workers; and nutritionists staff the unit during weekdays.
Our CICU is a sixteen-bed open-model ICU for cardiovascular medical patients and a closed ICU model for cardiac surgery patients. Resident house staff, advanced practice providers, cardiology fellows, and cardiology attending physicians lead and coordinate the daily management of cardiac medical patients. Critical care medicine consultation is provided at the discretion of the attending cardiologist by medical intensivists. Cardiac surgery patients are managed in a closed-unit model, with 24-h in-house cardiac surgery advanced practice providers, daily multidisciplinary ICU rounds, and cardiac surgery attending nighttime coverage by phone.
Our cardiac intensivist program started in January 2018 providing coverage of cardiac medical ICU admissions. The attending cardiologist retains primary responsibility and decision making. The cardiac intensivist provides critical care consultation as requested when not the primary cardiology attending of record, roughly 50% of patients. CICU multidisciplinary rounds are led by a board-certified cardiac intensivist on weekdays with in-unit coverage from 07:00 to 17:00 with no weekend or night coverage. Weekday coverage by a cardiac intensivist occurs 3 of 4 weeks of the month; the remaining week is covered by a dedicated interventional cardiology attending performing CICU multidisciplinary rounds with critical care medicine consultation as deemed necessary. Night coverage is provided by cardiology fellows, resident physicians, and advanced practitioners with general and interventional cardiology attending coverage by phone. For the purposes of this study, admissions during cardiac intensivist weeks and the interventional cardiology week were grouped together.
Patients over 18 years of age consecutively admitted to the CICU between July 1, 2012 and June 30, 2019 were enrolled. Only index admissions to the CICU were included. Demographic data including age, sex, race, admission service (medicine or cardiac surgery), ICU length of stay, ICU mortality, Charlson-Deyo and Elixhauser comorbidity indices, location before admission, Acute Physiology and Chronic Health Evaluation III (APACHE III) scores were collected from the electronic medical records (EMR) data warehouse.18–20 APACHE III scores were available after 2016, for approximately 34% of patients. The location prior to ICU admission was grouped from the catheterization lab, operating room, pre-admission (direct admissions), step down/floor, and other ICUs (medical, neurological, surgical). The primary insurance type was grouped by Medicaid, Medicare, commercial insurance. We categorized the patient’s admission types as elective, emergency, trauma, and urgent. Trauma admissions were cleared by the trauma service prior to admission to their respective units. Urgent admissions represented unscheduled admissions not originating from the emergency department; for example, scheduled elective surgery with complications, pre-admissions from home or a doctor’s office, or transfers from other facilities. We also analyzed the partial/hybrid cardiac intensivist model. There is no literature consensus on specific ‘weekday,’ ‘weekend,’ and ‘night’ or ‘off-hours’ timing. For the purposes of this study, weekday day admission was defined Monday-Friday from 08:00–17:59 h, outlining weekday night hours from 18:00 to 07:59. Weekend night hours were described between Friday 18:00 to Monday 07:59 while weekend day-time was bounded by Saturday and Sunday 08:00–17:59.
Summary statistics were calculated for patients’ demographic and clinical variables. For categorical variables, proportions were calculated and for continuous variables following normal distribution, mean and standard deviation are reported. For continuous variables with skewed distribution, median and interquartile range (IQR) are reported.
The unadjusted effect of arrival day and time on ICU mortality was estimated by fitting a logistic regression model without adjusting for other factors. Secondly, a multiple logistic regression model was employed to evaluate the factors associated with ICU mortality. The covariates included in the model were determined a priori based on clinical relevance and past literature rather than preliminary univariate analysis or arbitrary threshold p-value. The independent variables included in the multivariable regression models were arrival day and time, age, sex, race, ethnicity, primary language, primary insurance type, hospital admission type, previous unit location, patient type, and comorbid conditions of cerebrovascular disease, heart failure, chronic lung disease, liver disease, diabetes, renal failure, and myocardial infarction. We also adjusted for age, Charlson-Deyo comorbidity score, and adoption of a cardiac intensivist model. A continuous variable representing the year of admission was introduced in the model to adjust for possible temporal trends in ICU mortality.
Inherent differences exist between medical and surgical ICU patients. Medical admissions often present with ST-elevation myocardial infarctions (STEMIs), heart failure, and cardiogenic shock, leaning on multiorgan support therapies such as mechanical circulatory devices, renal replacement, mechanical ventilation, and vasopressor medications. Surgical patients are usually admitted post-operatively, often from coronary artery bypasses, valve replacements, or extracorporeal membrane oxygenation (ECMO) placement. These include both routinely scheduled elective cases and emergent surgeries. Owing to these differences, we conducted subgroup analysis by primary service (surgical or medical). In the subgroup analysis, an identical model formulation was maintained with dependent and independent variables as in the main model.
To evaluate the factors associated with the ICU length of stay (LOS), a generalized linear model (GLM) with log link function and gamma distribution was estimated. We fitted a similar model for the subgroup of surgical and medical ICU patients. For ease of interpretation the estimates from the gamma model are presented as ratio of expected length of stay calculated as eβ, where β is the regression coefficient from the gamma model. The patients who died in the CICU or had missing admission or discharge time were excluded from the analysis of ICU LOS.
Less than 1% of the data was missing. The patients with missing data were removed from analysis using a pairwise deletion approach resulting in available case analysis, where cases were excluded from only operations in which data were missing on a variable that was required. All statistical analyses were done using SAS® version 9.4 and visualization were developed using ggplot2 in R.21,22
Results
Our study included 10,628 admissions to the CICU from July 1, 2012 and June 30, 2019, admitted both to the cardiac medical and cardiac surgical ICU services, 58% and 42%, respectively (Table 1). The mean age of the study patients was 65 years (SD: 14). The patients admitted in the weekend were relatively younger than those admitted in the weekday with mean age of 66 years for weekday-day and weekday-night admission, and 64 and 65 years for weekend-day and weekend-night admissions. The majority of the patients were male (64%), White (79%), non-Hispanic or Latino (94%), and most spoke English (96%). Medicare insurance was most common (54%) While available only for 34% of patients (from year 2016 to 2019), the average APACHE III score was 49 (SD: 21). Elixhauser score overall was 14.6 (SD: 12)) and age-adjusted comorbidity Charlson-Deyo score was 7.2 (SD:3.6)
Table 1.
Characteristics of the study population. The values are count and percentage unless otherwise noted.
| Variable | All (n = 10,628) | Monday-Friday 08:00–17:59 h (n = 3782) | Monday-Friday 18:00–07:59 h (n = 4423) | Saturday-Sunday 08:00–17:59 h (n = 1037) | Saturday-Sunday 18:00–07:59 h (n = 1386) | P-value1 |
|---|---|---|---|---|---|---|
|
| ||||||
| Age* | 65.81 (13.91) | 66.26 (13.92) | 66.12 (13.86) | 64.30 (14.49) | 64.71 (13.50) | <0.001 |
| Sex | 0.748 | |||||
| Female | 3840 (36.13) | 1383 (36.57) | 1575 (35.61) | 371 (35.78) | 511 (36.87) | |
| Male | 6788 (63.87) | 2399 (63.43) | 2848 (64.39) | 666 (64.22) | 875 (63.13) | |
| Race | 0.005 | |||||
| Black | 1757 (16.53) | 660 (17.45) | 680 (15.37) | 193 (18.61) | 224 (16.16) | |
| White | 8389 (78.93) | 2927 (77.39) | 3568 (80.67) | 796 (76.76) | 1098 (79.22) | |
| Other† | 482 (4.54) | 195 (5.16) | 175 (3.96) | 48 (4.63) | 64 (4.62) | |
| Ethnicity | 0.005 | |||||
| Hispanic or Latino | 297 (2.79) | 115 (3.04) | 120 (2.71) | 30 (2.89) | 32 (2.31) | |
| Non-Hispanic or Latino | 9977 (93.87) | 3532 (93.39) | 4185 (94.62) | 955 (92.09) | 1305 (94.16) | |
| Language | 0.039 | |||||
| English | 10,217 (96.13) | 3626 (95.88) | 4263 (96.38) | 984 (94.89) | 1344 (96.97) | |
| Other | 411 (3.87) | 156 (4.12) | 160 (3.62) | 53 (5.11) | 42 (3.03) | |
| Insurance Type | <0.001 | |||||
| Commercial Insurance | 4591 (43.20) | 1582 (41.83) | 1889 (42.71) | 484 (46.67) | 636 (45.89) | |
| Medicaid | 301 (2.83) | 105 (2.78) | 113 (2.55) | 45 (4.34) | 38 (2.74) | |
| Medicare | 5736 (53.97) | 2095 (55.39) | 2421 (54.74) | 508 (48.99) | 712 (51.37) | |
| Medicine Service | 6194 (58.28) | 2465 (65.18) | 2052 (46.39) | 852 (82.16) | 825 (59.52) | <0.001 |
| Surgery Service | 4434 (41.72) | 1317 (34.82) | 2371 (53.61) | 185 (17.84) | 561 (40.48) | |
| Admission Type | <0.001 | |||||
| Elective | 2573 (24.21) | 500 (13.22) | 1713 (38.73) | 5 (0.48) | 355 (25.61) | |
| Emergency | 6569 (61.81) | 2681 (70.89) | 2086 (47.16) | 944 (91.03) | 858 (61.90) | |
| Trauma‡ | 36 (0.34) | 9 (0.24) | 15 (0.34) | 7 (0.68) | 5 (0.36) | |
| Urgent§ | 1450 (13.64) | 592 (15.65) | 609 (13.77) | 81 (7.81) | 168 (12.12) | |
| Previous Unit Location | <0.001 | |||||
| Catheterization Lab | 1214 (11.42) | 660 (17.45) | 384 (8.68) | 65 (6.27) | 105 (7.58) | |
| Operating Room | 3735 (35.14) | 1049 (27.74) | 2111 (47.73) | 100 (9.64) | 475 (34.27) | |
| Emergency | 3770 (35.47) | 1304 (34.48) | 1258 (28.44) | 628 (60.56) | 580 (41.85) | |
| Step Down/Floor | 1132 (10.65) | 531 (14.04) | 342 (7.73) | 155 (14.95) | 104 (7.50) | |
| Other ICUǁ | 184 (1.73) | 79 (2.09) | 64 (1.45) | 20 (1.93) | 21 (1.52) | |
| Pre-admission# | 553 (5.20) | 153 (4.05) | 242 (5.47) | 67 (6.46) | 91 (6.57) | |
| Other** | 40 (0.38) | 6 (0.16) | 22 (0.50) | 2 (0.19) | 10 (0.72) | |
| APACHE III Score *,†† | 48.86 (21.14) | 48.96 (20.95) | 49.40 (20.71) | 45.63 (22.67) | 49.05 (21.85) | 0.034 |
| Elixhauser Comorbidity Score* | 14.60 (12.47) | 15.70 (12.90) | 13.49 (11.94) | 15.66 (12.81) | 14.37 (12.38) | <0.001 |
| Charlson-Deyo Comorbidity Score* | 7.24 (3.60) | 7.53 (3.63) | 7.06 (3.51) | 7.17 (3.82) | 7.07 (3.56) | <0.001 |
| ICU Length of Stay (h)‡‡ | 35 (25, 65) | 37 (24, 70) | 34 (25, 58) | 43 (25, 71) | 34 (24, 63) | <0.001 |
| ICU Mortality | 674 (6.34) | 247 (6.53) | 230 (5.20) | 100 (9.64) | 97 (7.00) | <0.001 |
Chi-squared test implemented for categorical variables and Kruskal-Wallis test for the numerical variables.
P-values correspond to comparison among the 4 day-time groups.
mean and standard deviation.
Other racial groupings included Asian, American Indian, Native Hawaiian, other, unavailable, or declined.
Trauma admissions were cleared by the trauma service before admission to their respective services.
Urgent admissions represented unscheduled emergent admissions that did not originate in the emergency department; for example, scheduled elective surgery with complications, pre-admissions from home or a doctor’s office, or transfers from other facilities.
Other ICUs included medical, neurological, and surgical at our facility or as transfers from other institutions’ ICUs.
Pre-admissions included direct admissions.
Other previous unit locations included primarily gastroenterology endoscopy and obstetric triage units.
The APACHE III score was only available for 34% of the patients.
Median and interquartile range, in hours.
Overall, 674 patients (6.3% of the population) died in-hospital. The highest mortality was observed among patients admitted in weekend-day (9.6%), followed by weekend-night (7.0%), weekday-day (6.5%), and weekday-night (5.2%). In the bivariable analyses, without adjusting for covariates, compared to weekday-day admission, weekend-day admission was found to be associated with increased in odds of mortality by a factor of 1.53 (OR: 1.53, CI: 1.19–1.95) in overall population, by a factor of 1.85 (OR: 1.85, CI: 1.03–3.33), and 1.35 (OR:1.03–1.77) in surgical and medical patients respectively. In the multivariable analysis which adjusts for demographic and clinical factors, compared to weekday-day admission, weekend-day admission was found to be associated with increased odds of mortality by a factor of 1.33 (Adjusted OR (AOR): 1.33, CI:1.03–1.72) in the overall study population, and by a factor of 1.35 (AOR: 1.35, CI: 1.02–1.79) in medical patients. However, the admission day-time was not associated with mortality in the multivariable subgroup analysis of surgical patients.
Increasing age showed higher mortality, and the effect was more pronounced in the medical CICU admissions. Gender, race, ethnicity, and insurer were not independent CICU mortality risk factors. English-speaking patients had a lower mortality than non-English speaking, OR 0.46 (CI 0.31–0.667). Emergent versus planned admissions were not associated with an overall increased mortality, supported in subgroup analysis. Admissions from the catheterization lab were associated with a lower mortality compared to emergency department (ED) admissions (OR 0.57, CI 0.41–0.79). Operating room admissions from the surgical cohort also had a lower mortality over ED admissions (OR 0.338, CI 0.198–0.575). Admissions from other ICUs (including medical, neurological, surgical) compared to emergency department admissions had a significantly higher mortality (OR 4.62, CI 3.18 –6.71). Further, the comorbid conditions associated with increased mortality in the overall population include cerebrovascular disease, heart failure, liver disease, and renal failure. The comorbidity of heart failure was not associated with mortality in cardiac medical ICU admissions. A history of liver disease was strongly associated with cardiac surgical death, OR 12.43 (CI 4.99–30.95). Chronic lung disease, diabetes, and history of myocardial infarction demonstrated no difference in outcome. Implementation of an intensivist model did not change mortality overall, nor in the subgroup of medical and surgical patients.
Median ICU length of stay in the overall population was 35 (IQR: 25,65) hours. The longest ICU length of stay was observed in the patients admitted in the weekend-day (median: 43 h, IQR: 25–71), followed by weekday-day admission (median: 37 h, IQR: 24–70). In the multivariable analysis, the hospital admission day-time was not associated with ICU length of stay of the overall study population. However, compared to weekday-day admission, weekday-night admission was associated with 7% ICU longer length of stay in surgical patients, and 7% shorter ICU length of stay in medical patients. Previous unit location was significantly associated with ICU length of stay. In the overall study population, compared to patients admitted from ED, patients admitted from another ICU, catheterization lab, and step down/floor had 88%, 10%, and 17% longer ICU length of stay respectively. Similarly, history of cerebrovascular disease, heart failure, chronic lung conditions, liver disease, diabetes, renal failure, myocardial infarction were associated with longer ICU length of stay by 9%, 37%, 8%, 31%, 5%, 13%, respectively. The intensivist model was associated with a 12% longer LOS. In the surgical subgroup, compared to ED admissions, elective admission was associated with a 13% reduced LOS, and operating room admissions was associated with a 46% shorter LOS.
Discussion
Our data demonstrated an increased mortality for patients admitted on weekend days (Saturday-Sunday 08:00–17:59 h) compared to weekdays (Monday-Friday 08:00–17:59) for the entire cohort (Fig. 1). The increase in weekend day mortality was noted after adjustment for comorbidities. Specifically, mortality rate for all weekday admissions was 6.5% versus 9.6% for weekend days (OR 1.527, 95% CI 1.198–1.948, Table 2). Weekday and weekend night admissions were not associated with increased mortality set against weekday daytime admissions.
Fig. 1.
Distribution of patients and mortality by arrival day and time. This figure serves to present the distribution of all patients admitted to the cardiovascular intensive care unit (CICU) by arrival time (weekday [Monday-Friday] versus weekend [Saturday and Sunday] and day [08:00–17:59 h] versus night [18:00–07:59 h]), showing the columns of percent survived and deceased. Additionally, the right axis quantifies mortality percent of admission population with the red line emphasizing the increased mortality during weekend day-time (Saturday and Sunday 08:00–17:59 h) admission.
Table 2.
Effect of arrival day and time on ICU mortality. The results are from logistic regression models without accounting for covariates.
| Parameter | All (n = 10,628) | Subgroup Analysis | |
|---|---|---|---|
|
|
|||
| OR (95% CI) | Surgical ICU patients (n = 4434) OR (95% CI) | Medical ICU patients (n = 6194) OR (95% CI) | |
|
| |||
| Arrival Day and time | |||
| Monday-Friday 18:00–07:59 h | 0.785 (0.652, 0.945)* | 0.609 (0.427, 0.869)† | 1.051 (0.845, 1.308) |
| Saturday-Sunday 08:00–17:59 h | 1.527 (1.198, 1.948)† | 1.849 (1.027, 3.328)* | 1.350 (1.032, 1.767)* |
| Saturday-Sunday 18:00–07:59 h | 1.077 (0.844, 1.374) | 0.977 (0.606, 1.575) | 1.165 (0.877, 1.547) |
| Monday-Friday 08:00–17:59 h | Reference | ||
P<0.05.
P<0.01.
Cardiac surgery admissions’ mortality rate was not associated with admission day or time when matched for comorbidities (Table 3). Lower nighttime mortality appears driven by cardiac surgery admissions, while medicine admissions did not. The absence of increased weekend and weeknight mortality in the surgical group was surprising given their typically emergent nature and increased acuity. This could be explained by unknown confounders not included in our model or due to scheduling bias (elective surgeries are admitted during ‘night’ hours 05:00–06:00, but actually arrive to the CICU postoperatively during ‘day’ hours, Table 4). This scheduling bias could have skewed daytime versus nighttime analysis, but should not have impacted weekend admission comparisons as elective cases are routinely admitted on weekdays. Additionally, the lower severity of illness scores, as measured by a global acute and comorbidity index like APACHE III, noted in the weekend daytime admissions group argues against this hypothesis. Other well-known confounders described in the literature that explain the lack of weekend mortality include a closed unit ICU model and weekend multidisciplinary rounds.23
Table 3.
Factors associated with ICU mortality: results from the multiple logistic regression models.
| Parameter | All (n = 10,628) | Subgroup Analysis | |
|---|---|---|---|
|
|
|||
| Adjusted OR (95% CI) | Surgical ICU patients (n = 4434) Adjusted OR (95% CI) | Medical ICU patients (n = 6194) Adjusted OR (95% CI) | |
|
| |||
| Arrival day and time | |||
| Monday-Friday 18:00–07:59 h | 1.04 (0.853, 1.268) | 0.924 (0.617, 1.385) | 1.086 (0.864, 1.365) |
| Saturday-Sunday 08:00–17:59 h | 1.329 (1.026, 1.723)* | 1.191 (0.612, 2.318) | 1.354 (1.019, 1.799)* |
| Saturday-Sunday 18:00–07:59 h | 1.247 (0.963, 1.615) | 1.365 (0.805, 2.314) | 1.225 (0.908, 1.651) |
| Monday-Friday 08:00–17:59 h | Reference | ||
| Age † | 1.012 (1.003, 1.02)‡ | 1.011 (0.99, 1.032) | 1.014 (1.004, 1.024)‡ |
| Sex | |||
| Female | 0.94 (0.79, 1.118) | 1.265 (0.886, 1.804) | 0.846 (0.692, 1.035) |
| Male | Reference | ||
| Race | |||
| Black | 0.885 (0.703, 1.114) | 1.224 (0.759, 1.974) | 0.82 (0.63, 1.068) |
| Other § | 1.211 (0.821, 1.785) | 1.16 (0.5, 2.694) | 1.24 (0.795, 1.936) |
| White | Reference | ||
| Ethnicity | |||
| Hispanic or Latino | 0.709 (0.396, 1.269) | 1 (0.356, 2.808) | 0.587 (0.288, 1.197) |
| Unknown | 4.357 (3.123, 6.08)‡ | 6.448 (3.123, 13.314)‡ | 3.972 (2.712, 5.818)‡ |
| Non-Hispanic or Latino | Reference | ||
| Language | |||
| English | 0.457 (0.313, 0.666)‡ | 0.436 (0.201, 0.946)* | 0.472 (0.305, 0.73)* |
| Other Language | Reference | ||
| Insurance | |||
| Medicaid | 1.31 (0.804, 2.135) | 2.031 (0.786, 5.25) | 1.125 (0.623, 2.033) |
| Medicare | 1.006 (0.818, 1.239) | 1.163 (0.758, 1.785) | 1 (0.787, 1.271) |
| Commercial | Reference | ||
| Admission Type | |||
| Elective | 0.73 (0.5, 1.064) | 0.65 (0.403, 1.049) | 1.176 (0.617, 2.242) |
| Traumaǁ | 1.181 (0.439, 3.179) | 1.897 (0.411, 8.768) | 0.694 (0.158, 3.047) |
| Urgent# | 0.752 (0.536, 1.056) | 0.608 (0.319, 1.158) | 0.832 (0.553, 1.25) |
| Emergency | Reference | ||
| Previous Unit Location | |||
| Catheterization Lab | 0.569 (0.408, 0.793)‡ | 0.513 (0.361, 0.729)‡ | |
| Operating room | 0.27 (0.179, 0.407)‡ | 0.338 (0.198, 0.575)‡ | |
| Other** | 0.337 (0.044, 2.563) | 0.001 (0.001, 999.999) | 1.234 (0.159, 9.597) |
| Other ICU†† | 4.619 (3.182, 6.707)‡ | 11.301 (5.432, 23.512)‡ | 3.056 (1.91, 4.89)‡ |
| Pre-admission‡‡ | 1.026 (0.663, 1.588) | 1.1 (0.435, 2.786) | 0.993 (0.6, 1.644) |
| Step Down/Floor | 1.225 (0.968, 1.55) | 0.804 (0.388, 1.667) | 1.284 (0.997, 1.653) |
| Emergency | Reference | ||
| CICU Admitting Service | |||
| Medicine | 0.703 (0.535, 0.923)‡ | ||
| Surgery | Reference | ||
| Comorbid medical conditions§§ | |||
| Cerebrovascular Disease | 1.327 (1.094, 1.61)‡ | 1.689 (1.122, 2.543)* | 1.257 (1.005, 1.572)* |
| Heart Failure | 1.375 (1.134, 1.667)‡ | 2.072 (1.389, 3.092)‡ | 1.221 (0.975, 1.528) |
| Chronic Lung Conditions | 1.099 (0.913, 1.324) | 1.293 (0.868, 1.926) | 1.059 (0.855, 1.313) |
| Liver Disease | 2.635 (1.698, 4.088)‡ | 12.427 (4.989, 30.952)‡ | 1.945 (1.173, 3.225)‡ |
| Diabetes | 0.994 (0.827, 1.194) | 1.051 (0.708, 1.56) | 1.013 (0.82, 1.252) |
| Renal Failure | 1.395 (1.128, 1.724)‡ | 1.882 (1.149, 3.082)* | 1.354 (1.064, 1.721)* |
| Myocardial Infarction | 0.84 (0.703, 1.004) | 1.01 (0.68, 1.502) | 0.811 (0.66, 0.995)* |
| Charlson-Deyo Comorbidity Score | 1.051 (1.009, 1.095)* | 0.925 (0.814, 1.05) | 1.071 (1.025, 1.119)‡ |
| Admission year | 1.015 (0.956, 1.077) | 0.951 (0.84, 1.077) | 1.038 (0.969, 1.112) |
| Intensivist Model | 0.869 (0.646, 1.168) | 0.839 (0.456, 1.541) | 0.894 (0.634, 1.261) |
P<0.05.
Age is centered to mean.
P <0.01.
Other racial groupings included Asian, American Indian, Native Hawaiian, other, unavailable, or declined.
Trauma admissions were cleared by the trauma service before admission to their respective services.
Urgent admissions represented unscheduled emergent admissions that did not originate in the emergency department; for example, scheduled elective surgery with complications, pre-admissions from home or a doctor’s office, or transfers from other facilities.
Other previous unit locations included primarily gastroenterology endoscopy and obstetric triage units.
Other ICUs included medical, neurological, and surgical at our facility or as transfers from other institutions’ ICUs.
Pre-admissions included direct admissions.
The reference level for comorbid conditions is the absence of the specific comorbid condition.
Table 4.
Determinants of the CICU length of stay. The values are the exponential of the coefficients from the regression model, which is the ratio of expected length of stay.
| Parameter | All n = 9952 eβ (Ratio of Expected LOS) | Surgical ICU patients n = 4267 eβ (Ratio of Expected LOS) | Medical ICU patients n = 5685 eβ (Ratio of Expected LOS) |
|---|---|---|---|
|
| |||
| Arrival day and time | |||
| Monday-Friday 18:00–07:59 h | 0.98 (0.95, 1.02) | 1.07 (1.02, 1.12)* | 0.93 (0.88, 0.98)* |
| Saturday-Sunday 08:00–17:59 h | 0.99 (0.94, 1.05) | 0.98 (0.88, 1.09) | 0.99 (0.92, 1.06) |
| Saturday-Sunday 18:00–07:59 h | 0.96 (0.91, 1.01) | 0.96 (0.89, 1.02) | 0.97 (0.91, 1.04) |
| Monday-Friday 08:00–17:59 h | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Age | 1 0.00 (1.00, 1.00)* | 1 0.00 (1.00, 1.00)* | 1 0.00 (1.00, 1.00)* |
| Sex | |||
| Female | 1 (0.97, 1.04) | 1.07 (1.02, 1.11)* | 0.97 (0.93, 1.02) |
| Male | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Race | |||
| Black | 0.97 (0.93, 1.01) | 0.95 (0.89, 1.01) | 0.96 (0.90, 1.01) |
| Other† | 1.18 (1.08, 1.29)‡ | 1.13 (1.00, 1.29)* | 1.21 (1.07, 1.37)** |
| White | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Ethnicity | |||
| Hispanic or Latino | 1.01 (0.91, 1.13) | 1.05 (0.91, 1.21) | 0.93 (0.80, 1.08) |
| Unknown | 0.87 (0.78, 0.97)* | 1.03 (0.85, 1.26) | 0.84 (0.74, 0.96)* |
| Non-Hispanic or Latino | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Language | |||
| English | 0.98 (0.88, 1.08) | 0.91 (0.78, 1.05) | 1.00 (0.87, 1.15) |
| Other Language | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Insurance | |||
| Medicaid | 1.07 (0.97, 1.18) | 1.03 (0.87, 1.20) | 1.09 (0.96, 1.23) |
| Medicare | 0.97 (0.93, 1.01) | 1.00 (0.95, 1.05) | 0.95 (0.9, 1)* |
| Commercial | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Admission Type | |||
| Elective | 0.88 (0.83, 0.92)‡ | 0.87 (0.82, 0.92)‡ | 0.91 (0.79, 1.04) |
| Trauma§ | 0.82 (0.62, 1.08) | 0.57 (0.36, 0.90)* | 0.96 (0.67, 1.38) |
| Urgentǁ | 0.96 (0.90, 1.01) | 0.93 (0.87, 1.00) | 0.97 (0.89, 1.05) |
| Emergency | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Previous Unit Location | |||
| Catheterization Lab | 1.1 (1.04, 1.17)‡ | 1.09 (1.02, 1.16)* | |
| Operating room | 0.61 (0.57, 0.66)‡ | 0.54 (0.49, 0.59)‡ | |
| Other# | 0.72 (0.56, 0.93)* | 0.43 (0.33, 0.57)‡ | 1.12 (0.72, 1.73) |
| Other ICU** | 1.88 (1.64, 2.16)‡ | 1.43 (1.09, 1.87)* | 1.92 (1.62, 2.27)‡ |
| Pre-admission†† | 1.13 (1.03, 1.23)* | 1.05 (0.90, 1.23) | 1.12 (1.00, 1.25) |
| Step Down/Floor | 1.17 (1.10, 1.23)‡ | 0.76 (0.67, 0.87)‡ | 1.19 (1.11, 1.27)‡ |
| Emergency | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| CICU Admitting Service | |||
| Medicine | 0.68 (0.63, 0.73)‡ | ||
| Surgery | 1 [Reference] | ||
| Comorbid medical conditions‡‡ | |||
| Cerebrovascular Disease | 1.09 (1.05, 1.13)‡ | 1.04 (0.99, 1.09) | 1.14 (1.07, 1.2)‡ |
| Heart Failure | 1.37 (1.33, 1.42)‡ | 1.22 (1.17, 1.28)‡ | 1.50 (1.42, 1.58)‡ |
| Chronic Lung Conditions | 1.08 (1.04, 1.12)‡ | 1.02 (0.98, 1.07) | 1.12 (1.07, 1.18)‡ |
| Liver Disease | 1.31 (1.14, 1.51)‡ | 1.05 (0.8, 1.37) | 1.37 (1.16, 1.62)‡ |
| Diabetes | 1.05 (1.02, 1.09)* | 1.06 (1.02, 1.11)* | 1.05 (1.00, 1.11)* |
| Renal Failure | 1.13 (1.08, 1.19)‡ | 1.14 (1.07, 1.21)‡ | 1.13 (1.06, 1.2)‡ |
| Myocardial Infarction | 0.87 (0.84, 0.90)‡ | 1.00 (0.96, 1.05) | 0.82 (0.78, 0.86)‡ |
| Charlson-Deyo Comorbidity Score | 1.01 (0.99, 1.02)* | 1.00 (0.99, 1.02) | 1.01 (1.00, 1.03)* |
| Intensivist Model | 1.12 (1.07, 1.16)‡ | 1.17 (1.12, 1.23)‡ | 1.04 (0.98, 1.11) |
Notes: The values are exponential of the regression coefficient, which can be interpreted as a ratio expected ICU length of stay at a given level of a variable and the reference level of the variable. For example, the ratio of expected length of stay between female and male surgical patients is 1.07. This indicates that the female surgical ICU patients have 7% longer length of stay compared to the male surgical patients.
P<0.05.
Other racial groupings included Asian, American Indian, Native Hawaiian, other, unavailable, or declined.
P <0.001.
Trauma admissions were cleared by the trauma service before admission to their respective services.
Urgent admissions represented unscheduled emergent admissions that did not originate in the emergency department; for example, scheduled elective surgery with complications, pre-admissions from home or a doctor’s office, or transfers from other facilities.
Other previous unit locations included primarily gastroenterology endoscopy and obstetric triage units.
Other ICUs included medical, neurological, and surgical at our facility or as transfers from other institutions’ ICUs.
Pre-admissions included direct admissions.
The reference level for comorbid conditions is the absence of the specific comorbid condition.
Our open model cardiac intensivist coverage was implemented in January 2018. This subgroup analysis showed no difference in mortality by admission date or time. This data suggests that a partial cardiac intensivist model deployed in an open CICU (without weekend & night cardiac intensivist or multidisciplinary ICU rounds) had no difference in mortality. Our data is consistent with the general ICU literature, professional organization statements, and expert opinion describing intensivist-led, closed-model ICUs having lower ICU mortality without an increase in resource utilization.8,12,24–26 Other explanations for lack of mortality benefit include a smaller sample size (January 2018 to June 2019) and that no analysis of cardiac intensivist-only admissions were performed. This should not be interpreted as a lack of mortality benefit with cardiac intensivists, but rather a lack of benefit of the hybrid model. Further studies are needed.
ICU nurse and respiratory therapy staffing does not change over the weekend at our institution, so this unlikely affected our results. Other possible explanations for the increase in medical weekend day mortality was the lack of ICU pharmacist presence on weekend rounds.27 If this was true, the effect should have been noted both on medical and surgical admissions equally, which was not observed.
Admission time in this study was not analyzed as emergent versus non-emergent. This could have skewed our results to show an increased mortality on medical weekend admissions. As noted before, the severity of illness scoring (APACHE III) was lowest in the weekend day admissions, arguing against this effect. Additionally, surgical weekend admissions did not show increased mortality in the same timeframe.
This study is limited in several ways. Importantly, it is a retrospective electronic chart review. It is a single center study which may prevent generalization amongst other medical centers. Our analysis also did not account for holidays which may impact mortality. Further, the APACHE III severity of illness score was available only for 34% of the study patients and we did not adjust for APACHE III score in our models. The diagnosis coding in the EHR is done primarily for administrative purposes and is subject to coding error. This may have resulted in under or overestimation of the Elixhauser and Charlson-Deyo comorbid conditions used as covariates in the regression model. These results should be interpreted knowing medical admissions encompass a variety of admission types including emergent admissions in patients with shock, post-STEMI patients, routinely scheduled catheterization cases in high risk patients, all of which have variable acuity on presentation. Additionally, scheduled cardiac surgery procedures are an expectedly different cohort than emergent admissions and likely have additional unmeasured differences compared to patients admitted to the medical service or to the surgery service off-hours.
In summary, although multiple studies have shown conflicting mortality results regarding ICU admission day and time, none have consistently shown increased mortality during daytime weekday hours in a CICU.2–7 Rather, studies demonstrate a general increase in complications during weekends, nighttime, and holidays. This likely reflects a decrease in the level of staffing and institutional care processes during off hours versus weekday daytime admissions. Although tempting to attribute these findings to increased patient acuity, ours and other studies have adjusted for severity of illness. With these considerations in mind, the weekend effect seen in our medical admissions are most likely explained by different levels of physician staffing over weekends, the absence of weekend ICU multidisciplinary rounds, and the presence of an open rather than closed model ICU. Cardiovascular patients admitted to an open model CICU have an increased mortality if admitted over the weekend compared to weekdays. Physician staffing, ICU rounds, and patient care models should not have a significant variance between weekdays and weekends in cardiovascular ICUs. Further prospective, multi-center studies should be carried out to eliminate potential errors and biases. Closed CICU models with robust cardiac intensivist staffing should be examined for mortality benefit compared to other models.
Supplementary Material
Supplementary Figure 1. Distribution of patients admitted to the medical (orange) or surgical (green) cardiovascular intensive care unit (CICU) by admission hour. The large increase in surgical admissions between 05:00–07:00h was due to scheduled procedures formally admitted during these times but arriving to the CICU later in the day.
Acknowledgments
The authors would like to acknowledge invaluable assistance with database creation by Steven A DiSabatino, James T Laughery, Prathibha K Reddy, Ann Marie Lenoir, Deborah Moore, Ankhil Rao, Matthew Saponaro, and Marie L Cassalia.
Disclosures
The authors declare that there is no conflict of interest. KS is supported by an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health under grant number U54-GM104941 (PI: Binder-Macleod). The remaining authors have no financial disclosures or relationships with industry. No extramural funding was used to support this work.
Footnotes
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Supplementary materials
Supplementary material associated with this article can be found in the online version at doi:10.1016/j.hrtlng.2021.02.011.
References
- 1.Bell CM, Redelmeier DA. Mortality among patients admitted to hospitals on weekends as compared with weekdays. N Engl J Med. 2001;345(9):663–668. [DOI] [PubMed] [Google Scholar]
- 2.Arulkumaran N, Harrison DA, Brett SJ. Association between day and time of admission to critical care and acute hospital outcome for unplanned admissions to adult general critical care units: cohort study exploring the “weekend effect. Br J Anaesth. 2017;118(1):112–122. [DOI] [PubMed] [Google Scholar]
- 3.Kuijsten HAJM, Brinkman S, Meynaar IA, et al. Hospital mortality is associated with ICU admission time. Intensive Care Med. 2010;36(10):1765–1771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Cavallazzi R, Marik PE, Hirani A, Pachinburavan M, Vasu TS, Leiby BE. Association between time of admission to the ICU and mortality: a systematic review and metaanalysis. Chest. 2010;138(1):68–75. [DOI] [PubMed] [Google Scholar]
- 5.Bray BD, Steventon A. What have we learnt after 15 years of research into the weekend effect’? BMJ Qual Saf. 2017;26(8):607–610. [DOI] [PubMed] [Google Scholar]
- 6.Orsini J, Rajayer S, Ahmad N, et al. Effects of time and day of admission on the outcome of critically ill patients admitted to icu. J Community Hosp Intern Med Perspect. 2016;6(6):1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wunsch H, Mapstone J, Brady T, Hanks R, Rowan K. Hospital mortality associated with day and time of admission to intensive care units. Intensive Care Med. 2004;30(5):895–901. [DOI] [PubMed] [Google Scholar]
- 8.Pronovost PJ, Angus DC, Dorman T, Robinson KA, Dremsizov TT, Young TL. Physician staffing patterns and clinical outcomes in critically III patients: a systematic review. J Am Med Assoc. 2002;288(17):2151–2162. [DOI] [PubMed] [Google Scholar]
- 9.Khanna R, Wachsberg K, Marouni A, Feinglass J, Williams MV, Wayne DB. The association between night or weekend admission and hospitalization-relevant patient outcomes. J Hosp Med. 2011;6(1):10–14. [DOI] [PubMed] [Google Scholar]
- 10.Wilcox ME, Chong CAKY, Niven DJ, et al. Do intensivist staffing patterns influence hospital mortality following ICU admission? A systematic review and meta-analyses. Crit Care Med. 2013;41(10):2253–2274. [DOI] [PubMed] [Google Scholar]
- 11.Checkley W, Martin GS, Brown SM, et al. Structure, process, and annual ICU mortality across 69 centers: United States critical illness and injury trials group critical illness outcomes study. Crit Care Med. 2014;42(2):344–356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Morrow DA, Fang JC, Fintel DJ, et al. Evolution of critical care cardiology: transformation of the cardiovascular intensive care unit and the emerging need for new medical staffing and training models: a scientific statement from the American Heart Association. Circulation. 2012;126(11):1408–1428. [DOI] [PubMed] [Google Scholar]
- 13.Bohula EA, Katz JN, Van Diepen S, et al. Demographics, care patterns, and outcomes of patients admitted to cardiac intensive care units: the critical care cardiology trials network prospective north american multicenter registry of cardiac critical illness. JAMA Cardiol. 2019;4(9):928–935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Watson RA, Bohula EA, Gilliland TC, Sanchez PA, Berg DD, Morrow DA. Editor’s Choice-Prospective registry of cardiac critical illness in a modern tertiary care Cardiac Intensive Care Unit. Eur Hear J Acute Cardiovasc Care. 2019;8 (8):755–761. [DOI] [PubMed] [Google Scholar]
- 15.Jentzer JC, van Diepen S, Murphree DH, et al. Admission diagnosis and mortality risk prediction in a contemporary cardiac intensive care unit population. Am Heart J. 2020;224:57–64. 10.1016/j.ahj.2020.02.018. Epub 2020 Feb 28. PMID:32305724. [DOI] [PubMed] [Google Scholar]
- 16.Jentzer JC, Anavekar NS, Bennett C, et al. Derivation and validation of a novel cardiac intensive care unit admission risk score for mortality. J Am Heart Assoc. 2019;8:(17) e013675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Norton JM, Reddy PK, Subedi K, Fabrizio CA, Wimmer NJ, Urrutia LE. Utilization of an ICU severity of illness scoring system to triage patients with st-elevation myocardial infarction. J Intensive Care Med. 2020. 10.1177/0885066620928263. Epub ahead of print. PMID:32519573. [DOI] [PubMed] [Google Scholar]
- 18.Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45(6):613–619. [DOI] [PubMed] [Google Scholar]
- 19.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27. [DOI] [PubMed] [Google Scholar]
- 20.Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system: risk prediction of hospital mortality for critically ill hospitalized adults. Chest. 1991;100(6):1619–1636. [DOI] [PubMed] [Google Scholar]
- 21.SAS Institute Inc., Cary, NC U. SAS® 9.4. SAS Inst Inc, Cary, NC, USA 2017.22. R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria. 2017. [Google Scholar]
- 22.Kim MM, Barnato AE, Angus DC, Fleisher LF, Kahn JM. The effect of multidisciplinary care teams on intensive care unit mortality. Arch Intern Med. 2010;170(4):369–376. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.O’Malley RG, Olenchock B, Bohula-May E, et al. Organization and staffing practices in US cardiac intensive care units: a survey on behalf of the American heart association writing group on the evolution of critical care cardiology. Eur Hear J Acute Cardiovasc Care. 2013;2(1):3–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Amin P, Fox-Robichaud A, Divatia JV, et al. The Intensive care unit specialist: report from the task force of world federation of societies of intensive and critical care medicine. J Crit Care. 2016;35:223–228. [DOI] [PubMed] [Google Scholar]
- 25.Milstein A, Galvin RS, Delbanco SF, Salber P, Buck CR. Improving the safety of health care: the leapfrog initiative. Eff Clin Pract. 2000;3(6):313–316. [PubMed] [Google Scholar]
- 26.Lee H, Ryu K, Sohn Y, Kim J, Suh GY, Kim EY. Impact on patient outcomes of pharmacist participation in multidisciplinary critical care teams: a systematic review and meta-analysis. Crit Care Med. 2019;47(9):1243–1250. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Figure 1. Distribution of patients admitted to the medical (orange) or surgical (green) cardiovascular intensive care unit (CICU) by admission hour. The large increase in surgical admissions between 05:00–07:00h was due to scheduled procedures formally admitted during these times but arriving to the CICU later in the day.