Journal of Critical Care xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Critical Care journal homepage: www.jccjournal.org

Medical intensive care unit consults occurring within 48 hours of admission: A prospective study☆,☆☆,★ Rubin I. Cohen, MD a,⁎, Ann Eichorn, MS b, Caroline Motschwiller, BS a, Viera Laktikova, MD a, Grace La Torre, DO c, Nicole Ginsberg, MD a, Harry Steinberg, MD a a b c

Division of Pulmonary, Sleep and Critical Care Medicine, The Long Island Jewish Medical Center, The Hofstra North Shore–LIJ School of Medicine, New Hyde Park, NY Krasnoff Quality Management Institute, The North Shore-LIJ Health System, Manhasset, NY Department of Medicine, The Long Island Jewish Medical Center, The Hofstra North Shore–LIJ School of Medicine, New Hyde Park, NY

a r t i c l e

i n f o

a b s t r a c t

Keywords: MICU Unplanned transfers Functional status Elixhauser ICD-9

Rationale: Critical care consults requested shortly after admission could represent a triage error. This consult process has not been adequately assessed, and data are retrospective relying on discharge diagnoses. Objectives: The aims of this study were to identify reasons for medical Intensive care unit (MICU) consultations within 48 hours of admission and to detect differences between those accepted and those denied MICU admission. Methods: Data were prospectively collected including demographics, reason for consultation, Acute Physiology and Chronic Health Evaluation II score, Elixhauser comorbidity measure, functional status, need for assisted ventilation or vasopressor, presence of do-not-resuscitate (DNR) order, and whether a DNR order was obtained after MICU consultation. Results: Ninety-four percent of patients consulted were not initially evaluated in the emergency department, half of whom were accepted. Respiratory failure, sepsis, and alcohol withdrawal were the most frequent reasons for MICU transfers. Factors predicting MICU admission included respiratory illness, better baseline functional status, and less comorbidity, whereas DNR predicted rejection. We did not find differences in hospital mortality; but hospital length of stay was longer. Conclusions: Prospective examination of the consult process suggests that disease progression rather than triage error accounted for most unplanned transfers. Functional status and comorbidity predicted MICU admission rather than illness severity. Goals of care were not being discussed adequately. We did not detect differences in mortality although hospital length of stay was increased. © 2014 Elsevier Inc. All rights reserved.

1. Introduction

transferred to the ICU [1–4]. There are few data on why critical care consults are triggered shortly after hospital admission and why some patients are transferred to the ICU and others are not. To better identify whether an error in triage had occurred, it is important to assess the process from the time the ICU consult is requested rather than subsequent to patients' arrival to the ICU. We therefore undertook a 6-month prospective study to assess all patients admitted from the ED to the general medical ward (GMW) for whom a medical ICU (MICU) consult was requested within 48 hours of admission. Our primary aims were to identify the reasons for MICU consultations within 48 hours of admission and to detect any differences between those accepted and those denied MICU admission.

The transfer of patients to an intensive care unit (ICU) soon after hospital admission, also known as unanticipated or unplanned ICU transfer, may represent an initial triage error. On the other hand, such ICU admission may be due to unexpected clinical deterioration or due to development of a new problem. Studies assessing patients transferred to the ICU shortly after ward admission concluded that these transfers are associated with higher mortality and longer hospital stays compared with patients admitted directly from the emergency department (ED) and attributed these transfers to triage errors [1–4]. However, these studies were retrospective potentially limiting their conclusions. More importantly, previous studies did not investigate the ICU consult process; rather, they described patients already ☆ Support: None. ☆☆ Conflict of Interest: The authors declare that they have no conflict of interest with the work described herein. ★ This work was presented in part and in poster format at the 2012 ATS International Conference in San Francisco. ⁎ Corresponding author. E-mail address: [email protected] (R.I. Cohen).

2. Methods 2.1. Setting The study was conducted in a 600-bed adult academic hospital averaging 33600 annual admissions. The ED, with more than 68 000 annual visits, is the main route of patients' entry into the hospital. The MICU is an 18-bed closed model averaging 1300 annual admissions and is

http://dx.doi.org/10.1016/j.jcrc.2014.11.001 0883-9441/© 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Cohen RI, et al, Medical intensive care unit consults occurring within 48 hours of admission: A prospective study, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.11.001

2

R.I. Cohen et al. / Journal of Critical Care xxx (2014) xxx–xxx

continuously staffed by physicians certified in critical care medicine. The daytime MICU team consists of an attending, a pulmonary/critical care fellow, and 3 residents. During evenings and nights, there is an attending, 1 resident, and 1 physician assistant. Either a senior medical resident or a pulmonary/critical care fellow assigned to the MICU team evaluates all MICU consults. Medical ICU consults must also be evaluated by the attending, who is ultimately responsible for triage decisions. Consults must be completed within 45 minutes of being requested. The hospital does not have any step-down units, and medical patients who require mechanical ventilation (MV) are admitted to the MICU. A rapid response team consisting of medical house staff, not rotating in the MICU, a critical care nurse, and other supporting services is available at all times. A rapid response call does not necessarily result in MICU admission, and a critical care consult must be requested by the team caring for the patient. In a prior study, we determined that 21% of all rapid response calls are accepted to the MICU [5]. The study was approved by the health system's institutional review board that waived the need for informed consent. 2.2. Data collection Data were prospectively collected in “real time” by investigators in 2 ways: (1) using a standardized form and (2) reviewing medical records. We collected data from September to March to avoid changeovers in care teams, which occur at the academic end of year [6]. We included all patients admitted from the ED to GMW on whom MICU consult was requested within 48 hours of admissions. Should a patient be admitted to the GMW but remained in an ED holding unit while awaiting a ward bed, this patient counted as a consult had a MICU consult been requested within 48 hours of admission. We excluded critical care consults to the cardiac or surgical critical care units, patients transferred from another hospital, patients transferred from nonmedical ward units, and those who were planned MICU transfers after invasive procedures. (1) Data collection form: Collected data included demographics, reason for MICU consultation, elements necessary for an Acute Physiology and Chronic Health Evaluation II (APACHE II) score, the Elixhauser comorbidity measure [7], the presence of MV or noninvasive ventilation (NIV), the need for an intravenous vasopressor, the presence of a do not resuscitate (DNR) order at time of MICU consultation, and whether a DNR order was obtained after MICU consultation and the final triage decision. We also assessed patients' functional status by the previously validated Modified Rankin Score (MRS) [8–10]. The MRS is a commonly used scale for assessing the degree of disability and dependence. It ranges from 0 to 5, with lower scores indicating better performance, and scores of 4 or higher indicating severe disability and patients who are essentially bed bound. The MRS was assessed at 2 periods. The first MRS was the investigators' best determination of baseline functional status, defined as the patient's functional status before the acute illness that led to the present hospitalization. Prehospital functional status was defined by the patient (if able to communicate) or their family. Nursing assessment at hospital admission and the physician's history and physical were also relied on to determine baseline functional status. The second MRS was assessed by the investigators at the time of MICU consultation. Hospital length of stay (LOS) and discharge disposition were obtained using an administrative database (Premier Clinical Advisor, Charlotte, NC). (2) Medical record review: We considered that a patient's journey may not be fully captured by our data collection sheet. To more effectively assess events leading to MICU consultation, 2 clinicians (LV, GLT) not at all involved in the data collection described above, independently reviewed computerized medical records of each patient on whom MICU was requested within 48 hours of admission.

(3) International Classification of Diseases, Ninth Revision (ICD-9) diagnoses: We obtained ICD-9 discharge diagnoses and used the Clinical Classifications Software (CCS) for ICD-9-CM to further categorize the data [11]. Sponsored by the Agency for Healthcare Research and Quality, CCS is a database and software tool developed as part of the Healthcare Cost and Utilization Project. The CCS collapses ICD-9 codes into a smaller number of clinically meaningful categories, making it potentially more useful for presenting descriptive statistics than are individual ICD-9-CM codes. 3. Statistical analyses To improve discriminatory power, some continuous variables (APACHE II, MRS, and Elixhauser scores) were evaluated as categorical variables. The MRS scores were categorized as reflecting a level of disability that was low, medium, or high. For the APACHE II and Elixhauser scores, we created dichotomous variables indicating whether the patient's score was in the upper or lower half of the distribution of scores for all patients in the study. The median score for each variable (APACHE, 15; Elixhauser, 10) was used to divide the patients into these 2 categories. The κ statistic was used to determine interobserver reliability for the MRS. All variables were assessed with frequency distributions and evaluated for association with the outcome variables of triage decision and mortality with t tests or nonparametric equivalent (continuous variables) or χ 2 tests (categorical variables). A logistic regression model was constructed using the dichotomous outcome of MICU admission using a stepwise selection procedure run both backward and forward to evaluate reproducibility. Variables in the final model were those significant at the P = .05 level. Variables including sex, age, DNR status, MRS, Apache II, and Elixhauser comorbidity scores were also useful in measuring aspects of the patient's condition that would be considered in making the triage decision. We included these variables in the multivariate analysis. Data analyses were conducted using SAS v.9.13 (Cary, NC). 4. Results One hundred thirty-four MICU consults were requested within 48 hours of admission to the GMW, and 49% of these patients were accepted to the MICU (Fig. 1). The mean age of the admitted group was 62 ± 21 years, and 54% were male. Most of the patients (82%) came from home, and 18% were from long-term care facilities. Most (59%) of all MICU consults not requested by the ED (that is requested after admission) occurred within 48 hours, and approximately half of these patients were accepted. Fig. 1 shows that consults within 48 hours of GMW admission constituted almost 15% (134 of 917) of all MICU consults and 13% (66 of 521) of MICU admissions within the 6-month study period. 4.1. MICU consults requested within 48 hours of ward admission: accepted vs denied There was no significant difference in age or sex between those accepted or rejected. The most common stated reason for denying MICU admission was that the patient was not judged to be critical enough by the MICU team. When we examined diagnoses that led to MICU consult, only diagnosis that involved the respiratory system was significantly associated with MICU admission. Indeed, 58% of those with respiratory issues were accepted, whereas only 14% were rejected. Only 42% of those without the diagnosis of a respiratory problem were accepted (P = .04). Another factor that proved to be significant for acceptance or denial to the MICU was the presence of a DNR order. Those with DNR order were less likely to be admitted to the MICU (2 patients admitted vs 11 rejected; P b .01). Eight patients had a DNR order in place at the time of MICU consultation; the other 5 received a DNR order after discussions with the family by the MICU team (Table 1).

Please cite this article as: Cohen RI, et al, Medical intensive care unit consults occurring within 48 hours of admission: A prospective study, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.11.001

R.I. Cohen et al. / Journal of Critical Care xxx (2014) xxx–xxx

3

Table 2 Results of multivariable logistic regression model for MICU acceptance predictors for consults requested with 48 hours Effect: accept vs reject

OR

95% confidence limits

P

MRS high at MICU consult Elixhauser high Presence of DNR MRS high prehospital

13.045 0.303 0.172 0.132

4.733 0.128 0.030 0.042

b.001 .007 .047 .001

35.950 0.414 0.975 0.414

MRS high indicates severe disability. High Elixhauser score indicates the presence of more comorbidities. Because all patients requiring MV are admitted to the MICU, MV was not included in the model.

Fig. 1. Flowchart illustrating all MICU consults and outcomes (accepted/denied) within the 6-month study period.

APACHE II scores at the time of MICU consult were not different between those accepted or denied admission. The Elixhauser score, an index of comorbidity, was lower in those accepted to MICU (P b .01). In fact, with each increasing quartile of the Elixhauser score, MICU acceptance rate decreased (Table 2). For instance, 32% of patients with the highest score (most comorbidities) were accepted, whereas 60% in those with the lowest score (fewest comorbidities) were accepted. We found that patients were less likely to be accepted to the MICU if their prehospital (baseline) functional status was judged to be poor. Seventy percent of those severely disabled were denied MICU admission (P b .02). However, when we assessed level of disability at the time of MICU consultation, those who were severely disabled were more likely to be admitted to the MICU (61% admitted vs 39% rejected; P b .0007). Interobserver reliability for the MRS was 0.76 (P b .005). We found no difference in hospital mortality between those accepted or rejected to the MICU (15.4% accepted vs 15.9% rejected). Furthermore, there was no difference in the discharge location (home or skilled

Table 1 Characteristics of MICU evaluations within 48 hours of hospital admission

Age, y, mean ± SD Male, % (SE) APACHE II % of patients below median % of patients above median Elixhauser score Below, median (%) Above, median (%) MRS (prehospital) Mild (%) Moderate (%) Severe (%) MRS (at MICU evaluation) Mild (%) Moderate (%) Severe (%) DNR order, no. Hospital LOS, d, median (IQR) Hospital mortality (%)

nursing facility) in survivors between the 2 groups. The hospital LOS expressed as median [interquartile range (IQR)] was longer in those admitted to the MICU, 10 [6,16] compared with 8 [6,13] days than for those who remained on the GMW (P = .04). Fig. 2 demonstrates results of multivariate analyses predicting likelihood of being accepted to the MICU based on the delineated variables. More comorbidities, poor functional status at baseline (prehospital), and presence of DNR were all associated with an increased likelihood of MICU rejection. Conversely, poor functional status at the time of MICU consult correlated with an increased likelihood of MICU acceptance. 4.2. MICU consults requested directly from the ED vs within 48 hours of GMW admission Four hundred seven patients were accepted to MICU directly from ED compared with 65 patients who were accepted to MICU within 48 hours of admission. We found no differences between the 2 groups in terms of severity of illness, functional status (prehospital and at time of MICU evaluation), age, or sex. We also did not find a difference in overall mortality between those admitted directly to the MICU from the ED vs those admitted within 48 hours from the GMW. We then assessed mortality in patients with the diagnosis of respiratory failure, the most frequent reason for MICU consult and admission. We did not find a difference in this mortality either (28.4% admitted directly vs 24.1% admitted within 48 hours; P = .6). The only difference was that patients who were admitted from GMW had longer hospital LOS compared with those admitted directly from the ED (median IQR): 10 [6,16] vs 6 [3,10] days, P b .001. Multivariate analysis indicated a higher likelihood of MICU acceptance from the ED for patients with APACHE II score above the median (odds ration [OR], 1.9; 1.1-3.3). Those who were severely disabled at time of MICU 80 70

Admitted (n = 65)

Denied (n = 69)

P

59 ± 21 49 (5.9)

65 ± 19 51 (5.9)

NS NS

48 (6.2) 49 (6.0)

52 (6.2) 51 (6.0)

NS NS

63 (5.9) 35 (5.7)

37 (5.9) 65 (5.7)

.001 .001

56 (6.0) 50 (8.3) 30 (8.3)

44 (6.0) 50 (8.3) 70 (8.3)

NS NS .001

20

20 (17.9) 32 (6.6) 61 (5.5) 2 10 (6-16) 15.4 (4.4)

80 (17.9) 68 (6.6) 39 (5.5) 11 8 (6-13) 15.9 (4.4)

.001 .001 .001 .01 .04 NS

0

60 50 40

% admitted % rejected

30

10

Q-1

Q-2

Q-3

Q-4

Elixhauser score by quartile (Q) Fig. 2. Bar graph illustrating the Elixhauser scores in quartiles vs MICU admission. As comorbidities increased (higher score), the likelihood of MICU admission decreased (P b .01).

Please cite this article as: Cohen RI, et al, Medical intensive care unit consults occurring within 48 hours of admission: A prospective study, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.11.001

4

R.I. Cohen et al. / Journal of Critical Care xxx (2014) xxx–xxx

evaluation were less likely to be admitted from the ED (OR, 0.44; 0.24-0.85).

4.3. Medical record review Most MICU consults were on patients with respiratory issues. A third of these patients had chronic obstructive pulmonary disease (COPD) and acute or acute-on-chronic pCO2 retention requiring either MV or NIV. The rest had worsening pneumonia or acute chest syndrome, and 1 patient had worsening hemoptysis (Table 3). Patients with alcohol withdrawal (8% of consults) were another group with high MICU admission rate. These patients were initially treated in the GMW, but became too unstable or too difficult to manage, and 8 of the 10 were accepted. Eleven consults were requested for sepsis and 4 were rejected. One patient was diagnosed in the ED with aspiration pneumonia, but in fact had cardiogenic pulmonary edema that resolved with the administration of a diuretic. One patient was DNR and 2 had very poor functional status and multiple comorbidities. The MICU team felt that none of these 4 patients would benefit from MICU transfer. Of the 7 accepted patients with the diagnosis of sepsis, only one was initially evaluated in the ED with hypotension but responded to fluids and subsequently rejected by the MICU team. The other 6 patients were judged to be clinically stable in the ED and had normal blood pressure and normal or slightly elevated serum lactate levels. However, in the ED, 4 of these 6 patients had a white blood cell count N 15 000 × 103 μL, fever N 38.5°C, tachycardia (heart rate N 100 beats/min), and tachypnea (respiratory rate N 20 beats/min), and required supplemental oxygen. Within

Table 3 Diagnostic categories and outcomes by chart review Diagnostic categories

Denied, %

Admitted, %

Total, no. (%)

Hypotension (not due to sepsis)a Hypertension b Respiratory c Acute renal failure/electrolyte abnormalities d Infection/sepsis e Neurological f Anemia with presumed bleeding from any source g Overdose (mostly alcohol withdrawal) h Miscellaneous (abdominal pain, elevated amylase, weakness, perceived need for monitoring) i Cardiac arrest j

75 100 42 59 45 30 36

25 0 58 41 55 70 64

11 (8) 4 (3) 47 (35) 17 (13) 11 (8) 11 (8) 12 (9)

20 100

80 0

11(8) 6 (5)

0

100

4 (3) 134 (100)

a Patients denied admission if blood pressure improved with fluid administration or blood pressure was not low (error in measurement). Three patients were made DNR after MICU team discussed with family. b All responded to intravenous antihypertensive agents and denied MICU admission. c Most MICU consults were on patients with respiratory failure not associated with sepsis. A third of these developed either acute or acute on chronic pCO2 retention requiring mostly NIV. Patients were also accepted for worsening hemoptysis and deteriorating clinical status of acute chest syndrome, cystic fibrosis, community acquired, or aspiration pneumonias. Patients with respiratory issues denied MICU admission were those whose shortness of breath resolved when assessed by MICU team. Two COPD patients were examined by MICU team in the ED but were rejected. d Those with hyperkalemia and acute renal failure were likely to be admitted; those with hyponatremia were more likely rejected. e Those meeting criteria of sepsis (from nonrespiratory source) were generally admitted. f Accepted patients were those requiring frequent neurological checks for any reason (acute cerebrovascular accidents received tissue plasminogen activator, seizures not responding to initial management). g Medical ICU accepted those with presumed bleeding who were transfused but hemoglobin did not increase appropriately. h Treated with benzodiazepines on the medical wards but later became too difficult to manage; 80% were accepted. i None of these patients were deemed to require MICU admission. j All these patients died in the MICU.

24 hours of arrival to the GMW, these 4 patients required pressors for hypotension and 2 required MV. Patients admitted to GMW with anemia who were hemodynamically stable but whose hemoglobin level did not increase appropriately after blood transfusion were accepted to the MICU. Patients admitted for other etiologies but developed acute cerebrovascular accidents or seizures not responding to initial management were also accepted to the MICU, as were patients requiring neurological checks every 2 hours or more frequently. 4.4. Diagnosis classification by ICD-9 discharge codes We obtained the primary billing diagnosis and all secondary diagnoses. This group of 134 patients had 1604 diagnoses, an average of 12 diagnoses per patient. Further, we found that 31 patients had respiratory failure as a secondary diagnosis. When the diagnoses were categorized by CCS, 20 patients had a septicemia code as the primary, whereas we found 11 based on our chart review. We then reexamined the charts of the other 9 patients coded as septicemia by the billers/coders and found no clinical indication of sepsis. 5. Discussion In this single-center study, we did not detect a difference in hospital mortality between patients admitted to the MICU directly from the ED compared with those admitted within 48 hours. Further, we did not find a difference in hospital mortality when we specifically compared those with respiratory failure, the most common MICU admitting diagnosis between these 2 groups. In addition, we did not find a mortality difference between those accepted or denied MICU admission when the MICU consult was requested within 48 hours of GMW admission. We did find increased hospital LOS in those accepted to the MICU within 48 hours. Previous studies concentrated on patients already transferred to the ICU [1–3,12,13]. However, these previous studies did not consider the ICU triage process that might help clarify the cause of the transfer and undoubtedly affect who is transferred. Nevertheless, prior studies did conclude that such patients have higher in-hospital mortality and longer LOS compared with patients admitted to the ICU directly from the ED. The reasons for the difference in hospital mortality between our study and previous ones are not entirely clear, but may be due to study design and methods by which patients were included. Because we examined MICU consults, we did not assess cardiac or surgical conditions that would necessitate transfer to critical care units other than the MICU. Institutional factors such as the presence or lack of specialized units outside the ICU, teaching status, or staffing also play an important role. Indeed, after a review of available literature, Vlayen et al [14] concluded that it was impossible to estimate the incidence and outcomes of unplanned ICU transfers due to considerable heterogeneity among the 27 studies reviewed (of an initial pool of 83 studies). These authors found that the mean length of ICU stay ranged from 1.5 to 10.4 days, and mortality ranged between 0 and 58%. Previous studies relied on administrative databases and discharge diagnostic codes that may or may not reflect the condition they purportedly represent and could to be highly variable between both diagnoses and institutions. Moreover, the capture of diagnoses may be influenced by factors other than their existence in patients. These factors include physician documentation, accuracy of code assignment, and possible financial pressures [15]. These points are well illustrated by our results. Although it is true that our chart review determined reasons for MICU consult, rather than discharge diagnoses, we nevertheless note major differences between clinicians' review and billers coding practices. Coders assign a series of diagnoses from their review of the chart, and one is assigned as principal or primary and the condition responsible for the hospitalization. In addition, there can be a dozen or so secondary

Please cite this article as: Cohen RI, et al, Medical intensive care unit consults occurring within 48 hours of admission: A prospective study, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.11.001

R.I. Cohen et al. / Journal of Critical Care xxx (2014) xxx–xxx

diagnoses. It is entirely possible that what gets assigned by the coders as primary might not be the same as what a clinician would consider as primary. For example, when we examine specific diagnoses, we found that 31 patients were assigned respiratory failure not as primary, but as a secondary diagnosis, while our chart review found 35 patients with respiratory issues as primary reason for MICU consult (Table 3). Our chart review specifically separated those with respiratory failure due to sepsis from those with other etiologies (such as COPD). We also noted almost twice as many sepsis codes as clinical sepsis. When we reexamined the charts of such patients, we found that they all met systemic inflammatory response syndrome criteria but did not have sepsis clinically; however, systemic inflammatory response syndrome criteria were not intended as a standalone diagnosis of sepsis [16]. Our data indicate that in most consults requested from the GMW within 48 hours, the ED physicians had not considered an initial MICU evaluation to be necessary. In fact, only 8 of the 134 patients (6%) were initially assessed in the ED and rejected by the MICU team. Although critical care physicians were likely to accept more severely ill patients when the consult originated in the ED, severity of illness did not affect the likelihood of MICU acceptance when the consult originated from the GMW within 48 hours of admission. In these instances, critical care physicians placed more emphasis on patients' functional status and comorbidities, accepting those who were more functional at baseline (prehospital) but whose function deteriorated due to the acute illness. Indeed, those with more comorbidities and who were poorly functional at baseline (prehospital) were more likely to be denied MICU admission. This hints at the notion that MICU physicians may consider patients who are functional at baseline and those with less comorbidity to have a greater chance of meaningful recovery. This indicates that severity of illness and the number of comorbidities do not guarantee MICU transfer. In a retrospective cohort study of patients transferred to the MICU from non-ICU medical units at a university-affiliated hospital, Bapoje et al [1] determined that although most unplanned MICU admissions were due to either deteriorating illness or to development of a new one, 19% of such transfers were judged to involve errors in care. Furthermore, the authors determined that most such errors were potentially preventable and related to inappropriate admission triage. In another study, Delgado et al [17] retrospectively reviewed a database of a large health care provider and concluded that ED patients admitted with respiratory conditions (COPD, pneumonia), myocardial infarction, or sepsis were at modestly increased risk for unplanned ICU transfer and could benefit from improved ED triage, earlier intervention, or closer monitoring. We sought to ascertain whether patients who were admitted to the MICU within 48 hours should have been admitted directly from the ED. Our data suggest that rather than an initial triage error, health care practitioners may be underestimating illness trajectory or perhaps the decline was insidious. Multiple factors and at least 3 departments are involved including the ED, the MICU, and the GMW. Physicians appear to take into account multiple aspects when deciding whether to request a MICU consult (in the case of ED physicians) and to whether accept a patient (in the case of MICU physicians). In the ED, patients who were more ill (higher APACHE II score) and who were more functional at baseline (lower MRS score) were more likely to be admitted to the MICU if a consult was requested. This perhaps left those who were more functionally disabled and less ill. We did not study ED physician consult patterns; however, most patients (94%) on whom a MICU consult was requested within 48 hours of GMW admission did not undergo a MICU evaluation in the ED. Nevertheless, patients arrive at the ED at different stages of their illness and may be quite stable but deteriorate within 48 hours of admission. Alternatively, health care professional may not be proficient at assessing how each patient will fare. The possibility then arises as to whether the care delivered in the GMW was insufficient. We did not assess such care because this was not an objective of our study. Furthermore, other factors come into play after ED stabilization. These include the availability or lack of

5

GMW beds resulting in delay in transferring patients to the GMW as well as several handoffs between health care professionals. The impact, if any, of such factors requires further investigation. Our data indicate that most patients were clinically stable upon admission to the GMW and either deteriorated or developed a new problem. However, several caveats must be considered. Similar to other studies [1,17], most MICU consults requested within 48 hours of admission were due to respiratory instability. Most of our patients had a COPD exacerbation, and almost all had a prior history of pCO2 retention. They were admitted to the MICU because of deteriorating clinical status requiring assisted ventilation, usually NIV and occasionally MV. The second major reason for MICU transfers in our study was increased agitation secondary to alcohol withdrawal. The third category of note is patients who were eventually admitted to the MICU with sepsis but had a normal blood pressure in the ED, although most exhibited other abnormal vital signs. On the basis of ICD-9 discharge diagnoses, with their inherent deficiencies as noted above, there were 573 individual patients with alcohol intoxication, 995 patients with COPD, and 1187 patients with sepsis during the 6-month study period. Even if one third of these diagnoses were clinically relevant, this would still generate an inordinate amount of MICU consults. It appears that ED physicians are triaging properly, and most patients do not require MICU evaluation. In addition, our data suggest that patients admitted with both an acute COPD exacerbation and chronic pCO2 retention, or with evidence of clinical sepsis, should receive MICU consideration even if they appear stable. Delgado et al [16] suggested that developing mechanisms by which patients are routinely screened for meeting certain criteria might reduce the number of these so called unplanned MICU transfers. However, this would undoubtedly result in more MICU admissions, placing a larger burden on MICU bed availability. Alternatively, such patients could be admitted to a higher level of care such as a step-down unit with the hope of preventing or more effectively managing an acute decompensation. Our hospital does not have step-down units, and such units require more dedicated care and are therefore costly to initiate and maintain. On the other hand, these patients could be “flagged” for more concerted care on the GMW. We found that the presence of a DNR order impacted triage decisions. To our knowledge, no prior study assessed advanced directives in unplanned ICU transfers. We observed that patients with a DNR order were more likely to be denied MICU admission after MICU consult and discussion with the MICU team, 4 patients and families decided to forgo life-sustaining therapies, indicating that MICU consultations initiated end-of-life discussions. Interestingly, when we examined patients' charts, we found 7 families whose relative did not have a DNR order and was accepted to the MICU, decided to forgo further aggressive care within 72 hours of MICU admission. Our results imply that goals of care discussions are occurring either infrequently or too late on the GMW. In addition, our chart review indicated that palliative care consults were not requested. This suggests that GMW physicians caring for patients with advanced disease should have a better understanding of factors to identify patients for whom extension of life is not the priority. Our results may not be generalizable to other centers or to other types of ICUs. Because our hospital does not have a step-down unit, the possibility exists that less ill patients were admitted to the MICU with resultant lower mortality rate, making differences in mortality between groups more difficult to detect. The decision to admit patients to the MICU may be influenced by families or other clinicians. Our data show that the likelihood of accepting patients is partially based on the MICU team's perception of futility. Identifying such patients and communicating effectively are important strategies. 6. Conclusions In our single-center prospective study, we reexamined early MICU transfer process. Our data suggest that there was little triage error,

Please cite this article as: Cohen RI, et al, Medical intensive care unit consults occurring within 48 hours of admission: A prospective study, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.11.001

6

R.I. Cohen et al. / Journal of Critical Care xxx (2014) xxx–xxx

and most MICU transfers were due to disease progression. Indeed, 94% of these patients were not initially evaluated by the MICU team in the ED. When consulted, critical care physicians accepted half of these patients, taking into account functional status and comorbidity rather than severity of illness. We found that COPD with pCO2 retention, alcohol withdrawal, and clinical sepsis constituted most MICU transfers within 48 hours of admission. We also find that goals of care are not being discussed adequately and the MICU consult occasionally triggered such discussions. We did not detect any differences in hospital mortality between those accepted directly from the ED compared with those accepted within 48 hours, nor did we detect a mortality difference between those accepted or denied MICU admission when the consult was requested within 48 hours. However in both cases, the hospital LOS was increased. Authors' contribution RIC: project planning and implementation, data gathering and analysis, manuscript preparation, review, and submission. AE: data preparation, statistical analyses, manuscript review and revision. CM: project implementation, data gathering and analysis, abstract presentation at ATS conference, and manuscript revision. VL: data analysis. NG: data gathering and analysis, presentation of abstract at ATS conference, and manuscript revision. GL: data analysis. HS: data analysis, manuscript preparation, and revision. References [1] Bapoje SR, Gaudiani JL, Narayanan V, Albert RK. Unplanned transfers to a medical intensive care unit: causes and relationship to preventable errors in care. J Hosp Med 2011;6(2):68–72.

[2] Frost SA, Alexandrou E, Bogdanovski T, Salamonson Y, Parr MJ, Hillman KM. Unplanned admission to intensive care after emergency hospitalisation: risk factors and development of a nomogram for individualising risk. Resuscitation 2009;80 (2):224–30. [3] Keller AS, Kirkland LL, Rajasekaran SY, Cha S, Rady MY, Huddleston JM. Unplanned transfers to the intensive care unit: the role of the shock index. J Hosp Med 2010; 5(8):460–5. [4] Liu V, Kipnis P, Rizk NW, Escobar GJ. Adverse outcomes associated with delayed intensive care unit transfers in an integrated healthcare system. J Hosp Med 2012;7 (3):224–30. [5] Wang J, Lakticova V, McCabe AM, Amitrano B, Rosen L, Cohen RI. Critical care nurses' perception of time spent at rapid responses. Ann Am Thorac Soc 2013;10(3): 228–34. [6] Barach P. The July effect: fertile ground for systems improvement. Ann Intern Med 2011;155(5):331–2. [7] Van Walraven C, Austin P, Jennings A, Quan H, Forster AJ. A Modification of the Elixhauser Comorbidity measures into a point system for hospital death using administrative data. Med Care 2009;47(6):626–33. [8] Cohen RI, Eichorn A, Silver A. Admission decisions to a medical intensive care unit are based on functional status rather than severity of illness. A single center experience. Minerva Anestesiol 2012;78(11):1226–33. [9] Cohen RI, Lisker GN, Eichorn A, Multz AS, Silver A. The impact of do-not-resuscitate order on triage decisions to a medical intensive care unit. J Crit Care 2009;24(2): 311–5. [10] Aufderheide TP, Nichol G, Rea TD, Brown SP, Leroux BG, Pepe PE, et al. A trial of an impedance threshold device in out-of-hospital cardiac arrest. N Engl J Med 2011; 365(9):798–806. [11] Agency for Healthcare Research, Quality. The Healthcare Cost and Utilization Project. Clinical Classifications Software (CCS) for ICD-9-CM. [Internet]. Available from: http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp. [Accessed July 5, 2014]. [12] Mardini L, Lipes J, Jayaraman D. Adverse outcomes associated with delayed intensive care consultation in medical and surgical inpatients. J Crit Care 2012;27(6):688–93. [13] Frost SA, Davidson PM, Alexandrou E, Hunt L, Salamonson Y, Tam V, et al. Unplanned admission to the intensive care unit in the very elderly and risk of in-hospital mortality. Crit Care Resusc 2010;12(3):171–6. [14] Vlayen A, Verelst S, Bekkering GE, Schrooten W, Hellings J, Claes N. Incidence and preventability of adverse events requiring intensive care admission: a systematic review. J Eval Clin Pract 2012;18(2):485–97. [15] Rollins G. Improving the quality of quality measures: pilots show promising results from clinically enhanced administrative data. J AHIMA 2009;80(4):20–5. [16] Dellinger RP, Bone RC. To SIRS with love. Crit Care Med 1998;26(1):178–9. [17] Delgado MK, Liu V, Pines JM, Kipnis P, Gardner MN, Escobar GJ. Risk factors for unplanned transfer to intensive care within 24 hours of admission from the emergency department in an integrated healthcare system. J Hosp Med 2013;8(1):13–9.

Please cite this article as: Cohen RI, et al, Medical intensive care unit consults occurring within 48 hours of admission: A prospective study, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.11.001