Abstract
Objective
To better understand the mortality and notable characteristics of patients initially denied intensive care unit (ICU) admission that are later admitted on reconsultation.
Patients and Methods
We collected data regarding all adult inpatients (n=3725) who received one or more ICU consults at an academic tertiary care hospital medical center between January 1, 2018 and October 1, 2021. We compared patients who were initially denied ICU admission and later admitted on reconsultation (C2A1, n=144) with those who were admitted after the first consultation (C1A1, n=2286) and those denied at first consult and never later admitted (C1A0, n=1295).
Results
Ten percent of patients initially rejected by the ICU were later admitted on reconsultation. There was no significant difference in the adjusted hospital death odds ratios between C1A1 and C2A1 (0.67; 95% CI 0.43-1.01; P=.11). Assessing subgroups of the C2A1 population, we found that 8.2% (n=100) of full code patients were later admitted to the ICU on reconsultation vs 23.2% (n=40) of do not attempt resuscitation patients (P<.001); 7.6% (n=77) of patients initially consulted from the emergency department were later admitted to the ICU on reconsultation vs 15.1% (n=52) of patients initially consulted from an inpatient setting (P<.001).
Conclusion
In this cohort, we demonstrated that patients admitted on repeat ICU consultation have no significant difference in mortality compared with equivalent patients admitted after the first consultation. Understanding and further exploring the consequences of these ICU reconsultations is vital to developing optimal critical care triaging practices.
With need for critical care services rising as our population grows older and more medically complex, it is important for intensive care unit (ICU) providers to recognize patients that will most benefit from ICU admission.1,2 Although broad guidelines for intensive care unit admission, discharge, and triage have been previously established, consistent adherence with these recommendations is rare.3,4 The ICU acceptance rates vary widely between institutions, ranging anywhere between 15% and 72% of consultations received.5 Ideally, while the choice to admit should be based on objective data and institutional policy, external factors can affect the decision for ICU admission such as ICU bed availability,5, 6, 7, 8, 9, 10, 11 time of day,6 and the admitting provider’s experience level.12,13 Given the mortality benefit of timely critical care admission, accurate allocation of ICU resources is critical.5,14
Previously established features in commonly accepted and declined ICU consultations are revealing for the triaging process. Factors associated with acceptance to the ICU include a sudden increase in severity of illness, requiring active ICU treatment (rather than ICU observation), presence of hematological malignancy, vascular or hepatic disease involvement at triage, and having fewer comorbidities.5 Alternatively, ICU refusal is most often correlated with older age,5,12,15 multiple ICU triages during the current hospitalization,5 nature of the underlying disease process,12 presence of metastatic cancer,6 and worse functional status before hospitalization.5 These latter qualities illustrate a chronically ill segment of the population that seems more likely to be denied admission to the ICU.
Quickly providing ICU care to the appropriate patients is essential, as delayed ICU admission is associated with higher mortality rates and prolonged hospitalizations.16,17 After sudden deterioration, the survival benefit of ICU admission has been best demonstrated in the first 72 hours.18 A 4-hour to 6-hour delay in ICU transfer has been shown to significantly increase hospital length of stay and hospital mortality.19,20 Cardoso et al16 found that each hour of waiting when critical care services were indicated was independently associated with a 1.5% increased risk of ICU death.
Although the effects of ICU admission, ICU refusal, and delayed admission have been well-established, little literature exists regarding those who are initially denied ICU admission only to be accepted later on reconsultation. A 1-year, single-site study reported that 13% of patients initially denied ICU admission are subsequently admitted, with the most common reasons for reconsultation being sepsis, respiratory failure, and gastrointestinal bleeding.21 In this population of medical ICU patients accepted on reconsultation, the hospital mortality rate was 35.6% compared with the average medical ICU mortality rate of 31.9%.21 Furthermore, an Israeli, single-site 4-year analysis of mechanically ventilated patients initially denied ICU admission and subsequently accepted found that the odds of death were 3.6 times higher in this population compared with similar patients admitted on initial consultation.10 This study also found that these patients admitted later on reconsultation did not seem to benefit from the ICU admission,10 supporting previous findings regarding the diminished benefit of ICU care over time after initial deterioration.18
With around 1 in 8 patients that are initially denied ICU admission later being accepted, further understanding of the consequences of ICU admission decisions is vital.21 This study investigates if patients admitted to the ICU on reconsultation after initially being rejected have differential mortality compared with similar patients admitted at the first consultation.
Patients and Methods
Study Design, Population, and Setting
We performed a retrospective case-control study of all patients for whom the adult medical ICU was consulted at an academic tertiary care center from January 1, 2018 to October 1, 2021. We classified the study participants into 3 groups. The first group was patients for whom the ICU was consulted but who were denied and never later admitted (C1A0). The second group was patients who were accepted to the ICU on initial consultation (C1A1). The third group was patients who were initially denied by the ICU and were later accepted after reconsultation (C2A1). The primary outcome was all-cause hospital mortality. The institutional review board at our institution approved the study (VCU IRB–HM20023803).
Data Sources, Variables, and Measurements
We identified participants by querying all the medical ICU consultation orders in our electronic health record during the study period. We collected the location of the consulting physician, the timestamp of the consultation order, and the timestamp of the patient’s-level of care as ICU or arrival to ICU. If an individual had 2 medical ICU consult orders placed within a 3-hour time period, this was not considered a reconsultation; these instances were attributed to the requesting team inadvertently requesting the intial consultation twice and thefore were not of interest. On the basis of the collected information, patients were classified into study groups (C1A0, C1A1, and C2A1). We obtained the following variables: the timeline of hospitalization measured in quartiles (Q) (2018-Q1-2021-Q3), age (years), biological sex (female or male), race (Black vs non-Black), ethnicity (Hispanic vs non-Hispanic), admission source (emergency department [ED] vs inpatient vs other modes), the Charlson comorbidity index22 on the basis of ICD-10 present on admission codes only, and the daily calculated sequential organ failure assessment (SOFA) score.23
Bias
We adjusted for the severity of illness at the time of the first consultation (SOFA-C1 score) and the long-term comorbidity burden (POA-Charlson) on the basis of diagnoses present on admission only to address confounders. To minimize missing data from the SOFA scores, we calculated the SpO2/FiO2 (oxygen saturation/fraction of inspired oxygen) ratios for patients without arterial blood gas values, with missing laboratory values considered normal. We adjusted for the health care system’s adaptation to the COVID-19 pandemic and the different COVID-19 strains by adjusting for each quarter of the year as a categorical variable. Finally, we followed the Strengthening of The Reporting of Observational Studies in Epidemiology (STROBE) guidelines24 for the scientific communication of the results.
Statistical Methods
We obtained the largest cohort possible by selecting all eligible patients in the electronic health record without performing a priori power analysis because this is an epidemiological study. We used simple and multiple logistic regression models operating under the Bernoulli distribution for hospital mortality as a binary outcome. We conducted a sensitivity analysis to explore the possible confounding effect of the seasonality of the consultation.
Results
Participants and Descriptive Data
We retrospectively enrolled 3725 patients who received 1 or more ICU consults between January 1, 2018 and October 1, 2021. Among this cohort, 2286 patients (61.4%) were in the C1A1 group, 1295 (34.8%) were in the C1A0 group, and 144 (3.9%) were in the C2A1 group. The average age of the cohort was 57.7 years (SD, 16.1). The average SOFA-C1 score was 6.7 (SD, 2.5) and the average Charlson index was 2.9 (SD, 2.7). Of the total, 45.8% of patients (n=1707) identified as Black. 69.0% of initial consultations were from the ED (n=2570), whereas 24.2% were from the inpatient setting (n=902). Of the total, 86.7% of patients were full code (n=3228) at initial consultation, whereas 10.1% were do not attempt resuscitation (DNAR) patients (n=376). Additional details on the patient demographic characteristics and characteristics across the study groups can be found in Table 1.
Table 1.
Study Population
| Missing | Overall | C1A1 | C1A0 | C2A1 | |
|---|---|---|---|---|---|
| Patients, n | 3725 | 2286 | 1295 | 144 | |
| Sex, n (%) | |||||
| Male | 1777 (47.7) | 1034 (45.2) | 675 (52.1) | 68 (47.2) | |
| Female | 1663 (44.6) | 992 (43.4) | 620 (47.9) | 51 (35.4) | |
| Unidentified | 285 (7.7) | 260 (11.4) | 25 (17.4) | ||
| Race, n (%) | |||||
| Black | 0 | 1707 (45.8) | 989 (43.3) | 666 (51.4) | 52 (36.1) |
| Non-Black | 2018 (54.2) | 1297 (56.7) | 629 (48.6) | 92 (63.9) | |
| Ethnicity, n (%) | |||||
| Hispanic | 0 | 71 (1.9) | 40 (1.7) | 30 (2.3) | 1 (0.7) |
| Non-Hispanic | 3654 (98.1) | 2246 (98.3) | 1265 (97.7) | 143 (99.3) | |
| Admission source, n (%) | |||||
| ED | 0 | 2570 (69.0) | 1562 (68.3) | 931 (71.9) | 77 (53.5) |
| Inpt Transfer | 902 (24.2) | 558 (24.4) | 292 (22.5) | 52 (36.1) | |
| Other | 253 (6.8) | 166 (7.3) | 72 (5.6) | 15 (10.4) | |
| Code status, n (%) | |||||
| Full Code | 3228 (86.7) | 2004 (87.7) | 1124 (86.8) | 100 (69.4) | |
| DNAR | 376 (10.1) | 204 (8.9) | 132 (10.2) | 40 (27.8) | |
| Comfort | 0 | 121 (3.2) | 78 (3.4) | 39 (3.0) | 4 (2.8) |
| Age, mean (SD) | 0 | 57.7 (16.1) | 57.6 (15.9) | 57.8 (16.3) | 58.6 (15.8) |
| Charlson index, mean (SD) | 85 | 2.9 (2.7) | 2.9 (2.7) | 2.9 (2.7) | 2.9 (2.8) |
| SOFA-C1 score, mean (SD) | 19 | 6.7 (2.5) | 7.0 (2.6) | 6.0 (2.1) | 7.3 (2.8) |
| FiO2, mean (SD) | 19 | 46.0 (30.5) | 53.0 (32.2) | 33.2 (21.7) | 49.6 (32.1) |
| MAP, mean (SD) | 23 | 61.6 (14.5) | 59.0 (13.9) | 66.4 (14.5) | 58.5 (13.2) |
| Platelets, mean (SD) | 28 | 169.5 (113.4) | 163.3 (112.3) | 184.2 (114.8) | 138.1 (102.0) |
| Creatinine, mean (SD) | 26 | 2.4 (2.5) | 2.4 (2.6) | 2.2 (2.5) | 2.1 (1.7) |
| Bilirubin, mean (SD) | 455 | 2.8 (6.3) | 2.9 (6.5) | 2.3 (5.4) | 4.7 (9.0) |
| Year/quartile, n (%) | |||||
| 2018 / q1 | 0 | 250 (6.7) | 152 (6.6) | 88 (6.8) | 10 (6.9) |
| 2018 / q2 | 234 (6.3) | 149 (6.5) | 80 (6.2) | 5 (3.5) | |
| 2018 / q3 | 234 (6.3) | 140 (6.1) | 87 (6.7) | 7 (4.9) | |
| 2018 / q4 | 237 (6.4) | 144 (6.3) | 81 (6.3) | 12 (8.3) | |
| 2019 / q1 | 249 (6.7) | 154 (6.7) | 85 (6.6) | 10 (6.9) | |
| 2019 / q2 | 256 (6.9) | 154 (6.7) | 95 (7.3) | 7 (4.9) | |
| 2019 / q3 | 284 (7.6) | 173 (7.6) | 101 (7.8) | 10 (6.9) | |
| 2019 / q4 | 258 (6.9) | 159 (7.0) | 86 (6.6) | 13 (9.0) | |
| 2020 / q1 | 276 (7.4) | 164 (7.2) | 96 (7.4) | 16 (11.1) | |
| 2020 / q2 | 239 (6.4) | 139 (6.1) | 91 (7.0) | 9 (6.2) | |
| 2020 / q3 | 277 (7.4) | 172 (7.5) | 100 (7.7) | 5 (3.5) | |
| 2020 / q4 | 283 (7.6) | 168 (7.3) | 102 (7.9) | 13 (9.0) | |
| 2021 / q1 | 284 (7.6) | 189 (8.3) | 84 (6.5) | 11 (7.6) | |
| 2021 / q2 | 250 (6.7) | 152 (6.6) | 87 (6.7) | 11 (7.6) | |
| 2021 / q3 | 114 (3.1) | 77 (3.4) | 32 (2.5) | 5 (3.5) |
C1A0, (consults 1, ICU admissions 0) patients denied at initial ICU consult and never later admitted to the ICU; C1A1, (consults 1, ICU admissions 1) patients accepted at initial ICU consult; C2A1, (consults 2+, ICU admissions 1) patients denied at initial ICU consult, later admitted to the ICU on reconsultation; DNAR, do not attempt resuscitation; ED: emergency department; FiO2: fraction of inspired oxygen; Inpt: inpatient; MAP: mean arterial pressure; q, quartile; SD, standard deviation; SOFA-C1: sequential organ failure assessment score at initial ICU consultation.
Outcome Data
The adjusted odds ratio (OR) of all-cause hospital death of C1A1 and C2A1 compared with C1A0 was 1.67 (95% CI, 1.33-2.07; P<.001) and 2.42 (95% CI, 1.50-3.91, P<.001), shown in Figure 1. There was no significant difference between the OR of adjusted hospital death between C1A1 and C2A1 (0.67; 95% CI, 0.43-1.01; P=.11). No significant difference (0.17; 95% CI, 0.03-1.18; P=.07) was found on comparing the OR of adjusted hospital death in the ED subgroup of C1A1 (ED-C1A1) and C2A1 (ED-C2A1). Additional details regarding the comparative OR of hospital deaths can be found in Table 2.
Figure 1.
Effect of ICU rejection, acceptance, and reconsultation on mortality. Comparison of hospital mortalities adjusted for age, sex, demographic characteristics, time of year, the severity of illness at first consultation (sequential organ failure assessment at initial ICU consultation score), and comorbidities (Charlson index) between C1A0, C1A1, and C1A2. This comparison is also shown in Table 2. C1A0, (consults 1, ICU admissions 0) patients denied at initial ICU consult and never later admitted to the ICU; C1A1, (consults 1, ICU admissions 1) patients accepted at initial ICU consult; C2A1, (consults 2+, ICU admissions 1) patients denied at initial ICU consult, later admitted to the ICU on reconsultation; ICU, intensive care unit.
Table 2.
Multiple Logistic Regression of In-Hospital Death
| In-hospital death | Odds ratio | Standard Error |
t-value | P value | 95% CI | Significance | |
|---|---|---|---|---|---|---|---|
| C1A1 vs C1A0 | 1.67 | .185 | 4.53 | <0.001 | 1.332 | 2.065 | ∗∗∗ |
| C1A1 vs C2A1 | 0.69 | .160 | -1.62 | .105 | .434 | 1.082 | |
| C2A1 vs C1A0 | 2.42 | .591 | 3.62 | <0.001 | 1.499 | 3.905 | ∗∗∗ |
| Age (each year) | 1.03 | .004 | 7.55 | <0.001 | 1.02 | 1.034 | ∗∗∗ |
| Sex: Man | 1 | ||||||
| Woman | 0.94 | .093 | -0.66 | .509 | .772 | 1.137 | |
| Race: Black | 1 | ||||||
| Non-Black vs Black | 1.50 | .149 | 4.06 | <0.001 | 1.233 | 1.821 | ∗∗∗ |
| Hispanic | 1 | ||||||
| Non-Hispanic | 1.30 | .523 | 0.66 | .512 | .592 | 2.861 | |
| Charlson index (each point) | 1.20 | .02 | 10.47 | <0.001 | 1.156 | 1.235 | ∗∗∗ |
| SOFA-C1 score (each point) | 1.39 | .027 | 16.55 | <0.001 | 1.332 | 1.439 | ∗∗∗ |
| Constant | .001 | .001 | -13.63 | <0.001 | 0 | .003 | ∗∗∗ |
| Mean dependent variable | 0.189 | SD dependent variable | 0.391 |
| Pseudo r-squared | 0.178 | Number of observations | 3357 |
| Chi-square | 578.316 | Probability > chi2 | <0.001 |
| Akaike information criterion | 2690.133 | Bayesian information criterion | 2745.202 |
C1A0, (consults 1, ICU admissions 0) patients denied at initial ICU consult and never later admitted to the ICU; C1A1, (consults 1, ICU admissions 1) patients accepted at initial ICU consult; C2A1, (consults 2+, ICU admissions 1) patients denied at initial ICU consult, later admitted to the ICU on reconsultation; SOFA-C1, sequential organ failure assessment score at initial ICU consultation.
∗P<.1; ∗∗P<.05; ∗∗∗P<.01
The overall rate of initial ICU rejection was 38.6% (n=1439), with 10.0% (n=144) of these rejections later being admitted on reconsultation (C2A1). The rate of initial ICU rejection was 37.9% (n=1224) in full code patients vs 45.7% (n=172) in DNAR patients. Of the rejected patients, 8.2% (n=100) of full code patients were later admitted to the ICU on reconsultation vs 23.2% (n=40) of DNAR patients. The rate of initial ICU rejection was 39.2% (n=1008) in ED consults vs 38.1% (n=344) in those from an inpatient setting. Of these rejected patients, 7.6% (n=77) of ED patients were later admitted to the ICU on reconsultation vs 15.1% (n=52) of inpatient patients. Initial ICU consults for Black patients were denied 42.1% (n=718) of the time compared to 35.7% (n=721) of non-Black consults (P<.001). Of these rejected patients, 7.3% (n=52) of Black patients were later admitted to the ICU on reconsultation vs 12.7% (n=92) of non-Black patients. Figure 2 contains further details regarding initial rejection and subsequent reconsultation acceptances across different subgroups. Data regarding total consultations, ICU rejection rates, and reconsultation acceptance rates across the seasons throughout the duration of the study can be found in Table 1. The seasonality sensitivity analysis details of the dataset can be found in Figure 3.
Figure 2.
Rejection and reconsultation acceptance rates across subgroups. (A) Comparison of initial ICU rejection rates at first consult between full code and do not attempt resuscitation consults as well as between ED and inpatient transfer consults. (B) Comparison of rates of ICU admission on reconsultation after initial rejection (C2A1) between full code and do not attempt resuscitation patients (DNAR) as well as between emergency department and inpatient transfer patients. C2A1, patients denied at initial ICU consult but later admitted to the ICU on reconsultation (consults 2+, ICU admissions 1); DNAR, do not attempt resuscitation; ED, emergency department; ICU, intensive care unit.
Figure 3.
Effect of seasonality on reconsultation. Total number of ICU consults (blue bars = fall and winter/green bars = spring and summer), overall initial ICU rejection rate per quartile (blue line), and rate of ICU admission on reconsultation after initial rejection (C2A1) per quartile (red line) across the duration of the study (quartile 1, 2018 through quartile 2, 2021). Quartiles during the COVID-19 pandemic are noted by the grey horizontal bar titled “COVID-19”. C2A1, patients denied at initial ICU consult but later admitted to the ICU on reconsultation (consults 2+, ICU admissions 1); ICU, intensive care unit; Q, quartile.
Discussion
Key Results
The current study investigated mortality outcomes in all adult patients that received at least 1 ICU consultation over approximately 3-year period at a single center. We compared patients who were denied by the ICU and never later accepted (C1A0, Consults 1+, ICU admissions 0), patients who were accepted to the ICU at first consultation (C1A1, Consults 1, ICU admissions 1), and patients who were initially denied by the ICU and later accepted after at least 1 reconsultation (C2A1, Consults 2+, ICU admissions 1). When compared with C1A0, we found the adjusted hospital mortality in C1A1 and C2A1 was nearly twice as high. Therefore, those admitted to the ICU, either on initial consultation or later after additional consultations, had higher mortality than those declined and never later accepted to the ICU. However, when C1A1 and C2A1 were compared against each other, no significant difference was found. Thus, despite having eventual ICU need, be it because of incorrect initial evaluation or further deterioration after an initial consultation, being admitted on reconsultation did not significantly worsen mortality.
Interpretation
Our findings are in contrast to existing literature, which has established negative outcomes associated with delayed ICU admission like increased mortality13,16,17,19,20,25 and length of hospitalization.16,19 Other works have found that ICU benefits are most significant in the first 72 hours after initial deterioration, with only minimal positive effects afterwards.18 The delay to ICU experienced in the C2A1 population between first consultation and later admission does not seem to bear the same mortality consequences established in these previous studies. This may be because the delays in previous research were driven by ICU congestion13,16,17,19,20,25 and delays in ICU physician evaluation20 rather than the improper assessment of clinical trajectory or later exacerbating events. Therefore, these findings are reassuring to our ICU triaging practices and decisions to withhold ICU level care until it is appropriately indicated, even if ICU need is anticipated later in the hospitalization.
Key differences in the patient characteristics between C2A1 and C1A1 may help explain their similar mortalities despite C2A1’s delay to ICU admission. Compared with C1A1, C2A1 patients had an older average age, higher SOFA-C1 scores, a greater portion consults from the inpatient setting, and over 3 times more DNAR patients. C2A1 also appeared to have a higher incidence of comorbidities such as congestive heart failure, chronic pulmonary disorders, human immunodeficiency virus, metastatic cancer, and numerous others, though this data is difficult to clearly assess due to inconsistent admission documentation. These results support the Iapichino et al5 findings that patients requiring multiple ICU triages are more likely to be older, with lower performance status, and more severe illness compared with those who received only 1 evaluation.5 Patients with greater burdens of chronic illness may be more likley to get preemptive initial ICU consults, potentially based on smaller shifts in overall health compared with those with less chronic illness. Therefore, during this period of delayed ICU care between initial rejection and eventual admission, the more chronically ill C2A1 population may be closer to their baselines. This could be why C2A1’s delays do not demonstrate the mortality consequences that delayed ICU admissions have previously shown.16,18, 19, 20
Our findings show that code status of initially declined patients is an important predictor of later ICU need, especially if they are DNAR. Traits that have been associated with increased ICU rejection such as older age,5,15,20 greater pre-ICU dependency,5,6 higher severity of illness at time of triage,5,15 and perceived inability to benefit from ICU care6,15 are all seen more in the DNAR population compared to those that are full code. These features are also associated with a higher likelihood of needing multiple ICU consultations during a single hospitalization.5 Similar to previous works, our study found that DNAR status consults were initially denied more often than full code consults. However, after an initial rejection, almost 1 in 4 DNAR patients required reconsultation and eventual ICU admission, nearly 3 times more than what was observed in full code patients. Despite this, we do not see the reason to pursue more proactive admission practices in DNAR patients because admission at first consult was not associated with any significant improvement in mortality.
We also assessed differences between ED and inpatient ICU consults. Our study found that ED and inpatient ICU consultations were declined at near-equal rates. ED patients declined on initial ICU consultation were half as likely to be admitted on reconsultation compared with inpatient consults. This perhaps reflects the ED population’s greater likelihood of improving with continued care. However, there potentially could be a mortality consequence for those who do not improve. Although no significant difference in adjusted in-hospital mortality was found in the C2A1 population compared with C1A1, the ED portion of C2A1 had higher observed mortality. Specifically, ED patients rejected by the ICU at first consult and later accepted on reconsultation (ED-C2A1) had nearly 6 times higher adjusted hospital mortality when compared with ED patients accepted at first consult (ED-C1A1); however, this did not reach statistical significance. Our study is likely underpowered to investigate this question, and additional studies could be directed to further evaluate this trend.
The seasonal timing of consultation also correlated with rates of reconsultation. The sensitivity analysis indicated that over 3 years, the total number of ICU consultations appeared to be random across all 4 seasons. However, the proportion of patients admitted for reconsultation was higher in the fall (Q4) and winter (Q1) compared with spring (Q2) and summer (Q3). Previous studies have reported that late fall and early winter months correlate with periods of higher ICU census,26,27 which has been associated with longer delays in ICU admission and higher rates of denial.5,6,25,7, 8, 9,11,13,15,16,19 It is reasonable to assume that higher rates of initial rejection could contribute to more reconsultation admissions later in hospitalization. Additionally, previous research has shown critically ill patients admitted during the fall and winter months were more likely to develop viral and bacterial infections during their hospitalization.28 Other studies report that asthma and chronic obstructive pulmonary disorder exacerbations occur most often in the fall, whereas rates of sepsis and cardiac arrest are highest in the winter.29, 30, 31 These seasonal variations in disease process that could destabilize chronically ill populations may also explain the observed fluctuations in reconsultation admissions.
Significant trends were appreciated when comparing ICU denial and reconsultation acceptance rates across certain demographic characteristics. For example, Black patients were more likely to be denied after the first consult compared with non-Black patients. This underlines potential health care disparities that should be explored in future studies. However, the Black population was about half as likely to need ICU reconsultation and admission later in the hospitalization compared with the initially denied non-Black population. Differences across other demographic characteristics, including sex and Hispanic origin, were also assessed and did not show a significant difference in ICU denial or reconsultation rates, although the latter group was likely underpowered with only 71 total Hispanic-identifying patients in this study.
This research suggests numerous additional questions that should be further explored. Although reconsultation did not significantly affect mortality, it will be important to next address its effect on the length of hospitalization, which presumably would be extended owing to the delay in ICU admission. Also, as previously discussed, further evaluation of the ED reconsultation population is important to better understand the trend toward significant in-hospital mortality we observed. Similarly, exploring the observed higher rates of ICU rejection in the Black population is imperative and could help further identify potential health disparities in practice. Continued research in reconsultation will help to create objective universal ICU acceptance pathways and society recommendations.
Limitations
Notable limitations of our study include its retrospective nature. Furthermore, not all data could be retrieved from the electronic health record, leaving some information missing as discussed in the methods section. Additionally, we were not able to capture every reconsultation, as some reconsultations may have been verbally communicated or not entered as an electronic order request for other reasons. Additionally, a small portion of our reconsultation population was admitted to the ICU more than 5 days after their initial declined ICU consultation, potentially representing an entirely new process that led to this later decompensation. Addressing these limitations in forthcoming work will help solidify future findings.
Conclusion
ICU reconsultation after an initial rejection is not an uncommon occurrence. This study found that patients requiring ICU reconsultation in an academic institution did not experience a significant difference in mortality compared to patients accepted after the first consultation. Additional research is needed to further understand this reconsultation population, to improve ICU provider triage decision-making and provide optimal patient care.
Potential Competing Interests
The authors have no relevant conflicts of interest.
Acknowledgments
John Cyrus, MS, MA, from Health Sciences Library, Virginia Commonwealth University School of Medicine, Richmond, VA.
References
- 1.Needham D.M., Bronskill S.E., Calinawan J.R., Sibbald W.J., Pronovost P.J., Laupacis A. Projected incidence of mechanical ventilation in Ontario to 2026: preparing for the aging baby boomers. Crit Care Med. 2005;33(3):574–579. doi: 10.1097/01.ccm.0000155992.21174.31. [DOI] [PubMed] [Google Scholar]
- 2.Angus D.C., Kelley M.A., Schmitz R.J., White A., Popovich J. Current and projected workforce requirements for care of the critically III and patients with pulmonary disease: can we meet the requirements of an aging population? JAMA. 2000;284(21):2762–2770. doi: 10.1001/jama.284.21.2762. [DOI] [PubMed] [Google Scholar]
- 3.Task Force of the American College of Critical Care Medicine S of CCM Guidelines for intensive care unit admission, discharge, and triage. Task force of the American College of Critical Care Medicine, Society of Critical Care Medicine. Crit Care Med. 1999;27(3):633–638. [PubMed] [Google Scholar]
- 4.Azoulay E., Pochard F., Chevret S., et al. Compliance with triage to intensive care recommendations. Crit Care Med. 2001;29(11):2132–2136. doi: 10.1097/00003246-200111000-00014. [DOI] [PubMed] [Google Scholar]
- 5.Iapichino G., Corbella D., Minelli C., et al. Reasons for refusal of admission to intensive care and impact on mortality. Intensive Care Med. 2010;36(10):1772–1779. doi: 10.1007/s00134-010-1933-2. [DOI] [PubMed] [Google Scholar]
- 6.Garrouste-Orgeas M., Montuclard L., Timsit J.F., et al. Predictors of intensive care unit refusal in French intensive care units: a multiple-center study. Crit Care Med. 2005;33(4):750–755. doi: 10.1097/01.ccm.0000157752.26180.f1. [DOI] [PubMed] [Google Scholar]
- 7.Kim S.H., Chan C.W., Olivares M., Escobar G. ICU admission control: an empirical study of capacity allocation and its implication for patient outcomes. Manag Sci. 2015;61(1):19–38. [Google Scholar]
- 8.Robert R., Coudroy R., Ragot S., et al. Influence of ICU-bed availability on ICU admission decisions. Ann Intensive Care. 2015;5(1):55. doi: 10.1186/s13613-015-0099-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kim S.H., Chan C.W., Olivares M., Escobar G.J. Association among ICU congestion, ICU admission decision, and patient outcomes. Crit Care Med. 2016;44(10):1814–1821. doi: 10.1097/CCM.0000000000001850. [DOI] [PubMed] [Google Scholar]
- 10.Naser W., Schwartz N., Finkelstein R., Bisharat N. Outcome of mechanically ventilated patients initially denied admission to an intensive care unit and subsequently admitted. Eur J Intern Med. 2016;35:100–105. doi: 10.1016/j.ejim.2016.05.006. [DOI] [PubMed] [Google Scholar]
- 11.Mathews K.S., Durst M.S., Vargas-Torres C., Olson A.D., Mazumdar M., Richardson L.D. Effect of emergency department and ICU occupancy on admission decisions and outcomes for critically ill patients. Crit Care Med. 2018;46(5):720–727. doi: 10.1097/CCM.0000000000002993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Garrouste-Orgeas M., Montuclard L., Timsit J.F., Misset B., Christias M., Carlet J. Triaging patients to the ICU: a pilot study of factors influencing admission decisions and patient outcomes. Intensive Care Med. 2003;29(5):774–781. doi: 10.1007/s00134-003-1709-z. [DOI] [PubMed] [Google Scholar]
- 13.Louriz M., Abidi K., Akkaoui M., et al. Determinants and outcomes associated with decisions to deny or to delay intensive care unit admission in Morocco. Intensive Care Med. 2012;38(5):830–837. doi: 10.1007/s00134-012-2517-0. [DOI] [PubMed] [Google Scholar]
- 14.Edbrooke D.L., Minelli C., Mills G.H., et al. Implications of ICU triage decisions on patient mortality: a cost-effectiveness analysis. Crit Care. 2011;15(1):R56. doi: 10.1186/cc10029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Joynt G., Gomersall C., Tan P., Lee A., Cheng C.A., Wong E.L. Prospective evaluation of patients refused admission to an intensive care unit: triage, futility and outcome. Intensive Care Med. 2001;27(9):1459–1465. doi: 10.1007/s001340101041. [DOI] [PubMed] [Google Scholar]
- 16.Cardoso L.T.Q., Grion C.M.C., Matsuo T., et al. Impact of delayed admission to intensive care units on mortality of critically ill patients: a cohort study. Crit Care. 2011;15(1):R28. doi: 10.1186/cc9975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hung S.C., Kung C.T., Hung C.W., et al. Determining delayed admission to the intensive care unit for mechanically ventilated patients in the emergency department. Crit Care. 2014;18(4):1–9. doi: 10.1186/s13054-014-0485-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Simchen E., Sprung C.L., Galai N., et al. Survival of critically ill patients hospitalized in and out of intensive care units under paucity of intensive care unit beds. Crit Care Med. 2004;32(8):1654–1661. doi: 10.1097/01.ccm.0000133021.22188.35. [DOI] [PubMed] [Google Scholar]
- 19.Chalfin D.B., Trzeciak S., Likourezos A., Baumann B.M., Dellinger R.P., DELAY-ED study group Impact of delayed transfer of critically ill patients from the emergency department to the intensive care unit. Crit Care Med. 2007;35(6):1477–1483. doi: 10.1097/01.CCM.0000266585.74905.5A. [DOI] [PubMed] [Google Scholar]
- 20.Young M.P., Gooder V.J., McBride K., James B., Fisher E.S. Inpatient transfers to the intensive care unit: delays are associated with increased mortality and morbidity. J Gen Intern Med. 2003;18(2):77–83. doi: 10.1046/j.1525-1497.2003.20441.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Moran SL, Patel D, Chen L, et al. Characteristics of patients denied admission to the medical intensive care unit (MICU), then subsequently admitted. A94 ICU Util Outcomes. Published online 2014:A2184-A2184.
- 22.Sundararajan V., Henderson T., Perry C., Muggivan A., Quan H., Ghali W.A. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol. 2004;57(12):1288–1294. doi: 10.1016/j.jclinepi.2004.03.012. [DOI] [PubMed] [Google Scholar]
- 23.Vincent J.L., de Mendonca A., Cantraine F., et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on “sepsis-related problems” of the European Society of Intensive Care Medicine. Crit Care Med. 1998;26(11):1793–1800. doi: 10.1097/00003246-199811000-00016. [DOI] [PubMed] [Google Scholar]
- 24.von Elm E., Altman D.G., Egger M., et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Int J Surg. 2014;12(12):1495–1499. doi: 10.1016/j.ijsu.2014.07.013. [DOI] [PubMed] [Google Scholar]
- 25.Robert R., Reignier J., Tournoux-Facon C., et al. Refusal of intensive care unit admission due to a full unit: impact on mortality. Am J Respir Crit Care Med. 2012;185(10):1081–1087. doi: 10.1164/rccm.201104-0729OC. [DOI] [PubMed] [Google Scholar]
- 26.Garfield M., Ridley S., Kong A., Burns A., Blunt M., Gunning K. Seasonal variation in admission rates to intensive care units. Anaesthesia. 2001;56(12):1136–1140. doi: 10.1046/j.1365-2044.2001.01984.x. [DOI] [PubMed] [Google Scholar]
- 27.Lane J., Bhatnagar M., Lutrick K., et al. ICU resource limitations during peak seasonal influenza: results of a 2018 national feasibility study. Crit Care Explor. 2022;4(1) doi: 10.1097/CCE.0000000000000606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Miggins M., Hasan A., Hohmann S., et al. The potential influence of common viral infections diagnosed during hospitalization among critically ill patients in the United States. PLoS One. 2011;6(4) doi: 10.1371/journal.pone.0018890. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Danai P.A., Sinha S., Moss M., Haber M.J., Martin G.S. Seasonal variation in the epidemiology of sepsis. Crit Care Med. 2007;35(2):410–415. doi: 10.1097/01.CCM.0000253405.17038.43. [DOI] [PubMed] [Google Scholar]
- 30.Osborne M.L., Vollmer W.M., Buist A.S. Periodicity of asthma, emphysema, and chronic bronchitis in a northwest health maintenance organization. Chest. 1996;110(6):1458–1462. doi: 10.1378/chest.110.6.1458. [DOI] [PubMed] [Google Scholar]
- 31.Peckova M., Fahrenbruch C.E., Cobb L.A., Hallstrom A.P. Weekly and seasonal variation in the incidence of cardiac arrests. Am Heart J. 1999;137(3):512–515. doi: 10.1016/s0002-8703(99)70507-7. [DOI] [PubMed] [Google Scholar]