Article Text
PDF
Abstract
Purpose To evaluate the impact of implementation of a dedicated intensivist-led medical emergency team (IL-MET) on mortality in patients admitted to the intensive care unit (ICU).
Methods All adult ward admissions to the ICU between July 2002 and December 2009 were reviewed (n=1920) after excluding readmissions and admissions for <24 h. IL-MET hours were defined as 8:00–15:59 (Monday to Friday). The following periods were analysed: period 1: 1 July 2002–31 August 2004 (control); period 2: 1 September 2004–11 February 2007 (partial MET without dedicated intensivist); and period 3: 12 February 2007–31 December 2009 (hospital-wide IL-MET).
Results During all three periods, there were no significant differences in length of stay or mortality (IL-MET vs non-IL-MET hours, p>0.1 for all). On multivariate analysis, Acute Physiology and Chronic Health Evaluation (APACHE) II score and age were independently associated with mortality in all three periods (p<0.05 for all). During period 3, there was a non-significant trend towards decreased mortality if admitted during IL-MET hours (OR 0.73, 95% CI 0.51 to 1.03, p=0.08). During period 3, there was a non-significant trend towards decreased mortality if admitted during IL-MET hours (OR 0.73, 95% CI 0.51 to 1.03, p=0.08). However, this result likely reflects the observed increase in mortality during non-IL MET hours rather than improved mortality during IL-MET hours.
Conclusion In a single centre experience, implementation of an IL-MET did not reduce the rate of in-hospital death or lengths of stay.
- Critical care
- rapid response teams
- medical emergency team
- crisis management
- healthcare quality improvement
- implementation science
- hospital medicine
- decision support
- computerised
- transitions in care
- evidence-based medicine
- health services research
- patient safety
Statistics from Altmetric.com
- Critical care
- rapid response teams
- medical emergency team
- crisis management
- healthcare quality improvement
- implementation science
- hospital medicine
- decision support
- computerised
- transitions in care
- evidence-based medicine
- health services research
- patient safety
Introduction
Physicians are responsible for treating increasingly complex hospitalised patients. These patients often exhibit signs of physiological deterioration in the hours before cardiac arrest occurs.1 2 While cardiac arrest or ‘code’ teams have been around for decades, they often arrive late and/or are unsuccessful in more than 85% of cases, with survivors often at risk for significant hypoxic neurological insult.3 Multiple studies from Europe, North America and Australia have confirmed deficiencies in the way hospitals and standard models of care respond to acute illness on the ward.4–8 Because early detection of these warning signs may provide an opportunity for the prevention of in-hospital cardiopulmonary arrest and its associated poor clinical outcome, the use of rapid response systems (RRSs) has been promoted as a means of reduction of avoidable adverse events and in-hospital mortality. Recently, the Institute for Healthcare Improvement's One Hundred Thousand Lives Campaign has recommended that hospitals implement rapid response services or teams (RRTs) as one of six strategies to reduce preventable in-hospital deaths.9
The medical emergency team (MET) is the efferent arm of the RRS and is activated in response to simple, objective and reproducible criteria to provide, in a timely manner, the necessary resources to avert or reduce the probability of a poor clinical outcome for the ‘at-risk’ patient. Recent consensus guidelines differentiate MET teams from other RRTs in that they are physician led, whereas alternative models may be led by a nurse or respiratory therapist with or without physician consultation available.10 11 While data are inconclusive regarding the overall impact of implementation of RRSs on patient outcomes in acute care hospitals, emerging data are encouraging, and at present, RRSs continue to be broadly introduced.12–16
The University of Alberta Hospital initiated a hospital-wide dedicated intensivist-led MET (IL-MET) on 12 February 2007, operating during daytime hours (8:00–15:59) from Monday to Friday. While the MET runs 24 h a day, after these hours, it is led by the resident, nurse and respiratory therapist, who consult the on-call consultant intensivist. The aim of this study was to examine a dedicated IL-MET responding to rapid response calls/intensive care unit (ICU) consults from the medical and surgical wards in a large tertiary care centre and to assess the impact on clinical outcomes, most notably in-hospital mortality and length of stay.
Methods
The reporting of this study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline.17 The University of Alberta Health Research Ethics Board approved this study prior to commencement.
Study design and data collection
In this retrospective observational cohort study, we retrieved clinical data including age, sex, ICU admission time, diagnosis (International Classification of Diseases ninth revision (ICD9) codes), reason for ICU admission, source of ICU admission (emergency room, operating room, medical floor, surgical floor and other institutions) and Acute Physiology and Chronic Health Evaluation (APACHE) II score on admission. Our primary outcome measure was in-hospital mortality. Our secondary outcome measures were ICU mortality, and ICU and hospital lengths of stay.
Study population and setting
All ward source admissions to the General Systems Intensive Care Unit (GSICU) at the University of Alberta Hospital between 1 July 2002 and 31 December 2009 were reviewed (see figure 1). This is a 30-bed closed unit that admits medical and surgical (including trauma) patients and those who have had a solid organ transplant. Neurosurgical and cardiac surgical patients are admitted to separate dedicated units. There are three intensivist-led teams on at any given time and approximately 1600 patients per year are admitted to the GSICU. Inclusion criteria for this study were patients age 18 years or older, admission from the ward, duration of stay in ICU >24 h, and first ICU admission if several during the index hospitalisation. Of 9874 total admissions during the study period, 874 admissions were excluded as repeat admissions, 462 were for less than 24 h, 36 patients were <18 years old, 215 were missing key data (APACHE II score n=143 and source of admission n=72) and 6417 were excluded because their admission source was not from the ward (ie, emergency room, operating theatre, external referral). The remaining 1920 admissions were included in the study.
Summary of eligible ward admissions. Period 1 (1 July 2002–31 August 2004): no medical emergency team (MET) team. Period 2 (1 September 2004–11 February 2007): partial MET coverage (no dedicated intensivist). Period 3 (12 February 2007–31 December 2009): hospital-wide dedicated intensivist-led MET (IL-MET). IL-MET hours: Monday to Friday 8:00–15:59. Non-IL-MET hours: all other times out of IL-MET hours. ICU, intensive care unit.
Operational definitions
IL-MET hours
This was defined as any patient who was admitted to ICU between 8:00 and 15:59 on Monday to Friday (excluding statutory holidays). Patients admitted outside of these hours were considered non-IL-MET hour admissions. While MET could be activated out of these hours, it was primarily nurse/house staff driven with an available consultant intensivist on-call for review. For comparison, we divided up ward admission into three time periods: 1 July 2002–30 August 2004 (period 1: pre-introduction of the MET team); 1 September 2004–11 February 2007 (period 2: introduction of MET team covering part of the hospital without a dedicated intensivist); and 12 February 2007–31 December 2009 (period 3: introduction of IL-MET team covering all medical and surgical floors). Distributions of admissions included in this study are shown in figure 1.
MET activation
The MET is composed of an ICU resident and/or fellow, one ICU nurse and two ward-based respiratory therapists. During period 2 (limited MET coverage), the intensivist who was on intake in the ICU would be responsible for the MET along with all other patients admitted to the ICU. During period 3 (after 12 February 2007), a dedicated intensivist who was solely responsible for MET activity and ICU ward consults, who did not have any other responsibilities in the ICU, and was responsible for MET coverage within the entire hospital during previously defined hours.
Any member of the hospital staff could activate the MET. ‘Triggers’ for MET activation are based on several objective criteria that focus on changes to patients' clinical condition and acute physiology (ie, vital signs) and are outlined in table 1. Once the MET has been activated, the team is expected to respond within 15 min. The MET performs a rapid assessment, orders appropriate diagnostic tests and initiates treatment as necessary. The MET has medications, equipment and technology for acute resuscitation and endotracheal intubation, if necessary. Within 30 min, a decision is to be made on whether patients should be transferred to ICU for a higher level of support, or whether they can be safely managed on the ward.
Summary of criteria for activation of the medical emergency team (MET) system
Data sources, collection and storage
Data sources included the University of Alberta ICU-specific Minimal Data Set (MDS) database and hospital administrative databases. We extracted data on dates/time of ICU admission, primary diagnostic category, illness severity (ie, APACHE II score), mechanical ventilation, ICU and hospital lengths of stay, and vital status at ICU and hospital discharge.
Statistical analysis
Analysis was performed using Intercooled Stata Release 10 (Stata Corp, College Station, Texas, USA). In the event of missing data values, data were not replaced. Normally or near normally distributed variables were reported as means with SD and compared by Student t test and ANOVA test if appropriate. Non-normally distributed continuous data were reported as medians with IQR and compared using non-parametric tests (Wilcoxon rank sum and Kruskal–Wallis) where appropriate. Categorical variables were expressed as proportions and compared with the χ2 test. A customised multiple variable logistic regression model consisting of hospital mortality as a dependent variable and APACHE II score, age, use of mechanical ventilation, medical admission and IL-MET ward admission hours (reference was non-IL-MET hours) as independent variables. As this was an exploratory analysis, backwards logistic regression was performed. All statistical tests were two sided and p>0.05 was considered significant.
Results
Univariable analysis
Baseline characteristics, comorbid conditions and primary ICU diagnoses are shown in table 2. Statistical comparisons were made between IL-MET and non-IL-MET hours (p*) and between period 1 (no MET) and period 3 (hospital-wide MET, p**). Of the admissions prior to September 2004 (period 1), 143 (30%) occurred from 8:00 to 15:59 (MET hours) while 336 (70%) occurred out of IL-MET hours. Between September 2004 and February 2007 (period 2), 185 (29%) admissions occurred during IL-MET hours and 455 (71%) occurred out of IL-MET hours. Between February 2007 and December 2009 (period 3), 259 (32%) admissions occurred during IL-MET hours and 542 (68%) occurred out of IL-MET hours. On univariable analysis, there were no statistically significant differences in age or sex across all groups. Mean APACHE II scores were significantly higher in period 3 compared with period 1 (see figure 2, p=0.009). More patients in period 3 had two or more pre-existing comorbidities compared with period 1 (p**=0.02). For all groups, the most common primary ICU admission diagnosis was respiratory failure (greater than 40% of admissions). Compared with period 1, during period 3 more patients were admitted with a primary diagnosis of sepsis (17% vs 8%, p**<0.001) and fewer with respiratory failure (p**=0.007).
Baseline characteristics of the study patients at intensive care unit (ICU) admission
In-hospital mortality (%) and Acute Physiology and Chronic Health Evaluation (APACHE) II data for 1920 ward admissions between July 2002 and December 2009. Period 1 (1 July 2002–31 August 2004): no medical emergency team (MET) team. Period 2 (1 September 2004–11 February 2007): partial MET coverage (no dedicated intensivist). Period 3 (12 February 2007–31 December 2009): hospital-wide dedicated intensivist-led MET (IL-MET).
Patient outcomes are listed in table 3. During all three periods, there were no statistically significant differences in crude ICU or hospital mortality, ICU or hospital length of stay or requirement for mechanical ventilation (IL-MET vs non-IL-MET hours >0.1 for all). There was a non-significant trend towards decreased hospital mortality if admitted during IL-MET hours during period 3 (30.1% vs 35.9%, p=0.1). In overall comparisons between period 1 (pre-MET) and period 3 (hospital-wide IL-MET), there was a non-significant trend towards increased length of hospital stay during period 3 (25 (13–54) vs 29 (15–55) days, p=0.06).
Mortality, lengths of stay and necessity of mechanical ventilation differences between intensivist-led medical emergency team (IL-MET) hours and non-MET (non-IL-MET) hours during all three study periods
Figure 3 shows the absolute number of MET activations along with the MET activation rate (per 1000 admissions) between 2006 and 2009. By linear regression, there was a statistically significant increase in MET dose (number of activations per 1000 admissions) between 2006 and 2009 (p=0.012).
Total Number of medical emergency team (MET) activations (n) and MET activation rate (per 1000 admissions) between 2006 and 2009.
Multivariable analysis
Using logistical regression, a multivariable model was constructed from variables previously validated in the literature (age, APACHE II, comorbidity) as well as mechanical ventilation to determine if admission to ICU during MET hours for all three periods had any impact on patient mortality. These results are shown in table 4.
Multiple variable logistic regression analysis showing the association of in-hospital mortality with APACHE II, age, medical comorbidity, use of mechanical ventilation and admission during the intensivist-led MET (IL-MET) hours for all three study periods and from 1 January 2008 to 31 December 2009
In a further secondary exploratory analysis, for admissions during period 1 (pre-introduction of MET), APACHE II (per unit) (OR 1.09, 95% CI 1.06 to 1.13, p<0.001) and at least one comorbidity (OR 1.62, 95% CI 0.98 to 2.45, p=0.04) on ICU admission were independently associated with increased hospital mortality, while there was no association with admission to ICU during hours when the IL-MET would operate (OR 0.97, p=0.9). For admissions during period 2 (non-IL-MET), APACHE II (OR 1.10, 95% CI 1.07 to 1.13, p<0.001) and age (per year) (OR 1.02, 95% CI 1.00 to 1.03, p=0.002) independently predicted higher hospital mortality, while admission to ICU during IL-MET hours (OR 1.18, p=0.41) did not. During period 3 (hospital-wide IL-MET), APACHE II (OR 1.11, 95% CI 1.08 to 1.13, p<0.001), age (OR 1.01, 95% CI 1.00 to 1.02, p=0.008) and having one or more comorbidity (OR 1.43, 95% CI 1.01 to 2.02, p=0.04) were all independently predictive of increased mortality. During this period, there was a trend for decreased mortality if admitted during IL-MET hours (OR 0.73, 95% CI 0.51 to 1.03, p=0.08) compared with admission during non-MET hours. When including admissions only from January 2008 to December 2009, adjusted mortality was significantly lower if admitted during IL-MET hours (OR 0.57, 95% CI 0.38 to 0.87, p=0.01).
Discussion
We performed a retrospective observational cohort study of all adult ward source ICU admissions between July 2002 and December 2009 to evaluate the impact of a dedicated IL-MET on mortality, and ICU and hospital lengths of stay.
Key findings
We found that the implementation of a dedicated IL-MET was not associated with a statistically significant difference in overall hospital mortality. There were also no significant differences in ICU or hospital lengths of stay. In a secondary exploratory analysis, we found a lower adjusted mortality for patients admitted during IL-MET hours following the implementation of a hospital-wide dedicated IL-MET. However, this analysis likely reflects an increase in mortality during non-IL-MET hours rather than improved mortality during IL-MET hours. We also found that after February 2007, advanced age, severity of illness (quantified by APACHE II) and increased burden of pre-existing comorbid illness also independently influenced in-hospital mortality.
Limitations
Our study should be interpreted in the context of the following limitations. Like other studies, this was a ‘before–after’ retrospective cohort study using historical controls and evaluating a complex inter-disciplinary intervention; as such it may be prone to bias and confounding.18 19 We may not have fully adjusted for other quality improvement efforts that may have influenced study outcomes after the implementation of MET. We did not have data on the ‘do not resuscitate’ status for our study population, either on admission or established during admission, which may have impacted on our ability to detect a mortality benefit for patients selected for ICU admission and advanced life support. Moreover, we are not able to specifically comment on whether the decision to establish DNR status was affected by time of day and/or the presence of a dedicated intensivist. In addition, we did not have reliable estimates of rates of in-patient cardiopulmonary arrests during the study period. Furthermore, we were unable to adjust our analysis for any diurnal variation in MET calls. From 2004 to 2011, 41.3% (1322/3196) of all MET calls occurred between 8:00 and 15:59 (when a dedicated intensivist was present), 32.4% (1037/3196) occurred from 4:00 to 13:59 and 26.1% (837/3196) occurred from 12:00 to 7:59. Our database also does not enable adjustment for the duration of time a patient may have fulfilled criteria for MET activation prior to the response. Despite these limitations, our study was large and systematic in capturing all ICU admissions. We believe these data are relevant when considering the variation on models for MET implementation and that further data generated from randomised trials will be unlikely due to the complexity and absence of clinical equipoise. The only study to address an alternative study design was the Medical Emergency Response and Intervention Trial (MERIT), which used a large cluster model, randomly assigning participating hospitals to current standard of care compared with the hospital-wide introduction of a MET.12 Unfortunately, in part related to issues of study implementation and design, no significant differences in rates of cardiac arrest, unplanned ICU admission or unexpected death were found.
Comparison with prior literature
We believe our data are consistent with previous studies of rapid response teams.16 20 For example, in a large retrospective cohort study by Afessa and colleagues, which compared mortality for patients admitted to ICU during or outside of morning bedside rounds, the implementation of a RRT (ICU fellow led) was not found to impact the observed ICU mortality rates whether admission occurred during morning bedside rounds (8:00–11:00) or outside round time (13:00–6:00) (13.3% vs 13.5% p=0.90).21
While there are few data on the positive impact of MET on mortality, there are several single-centre before-and-after studies in the literature that primarily commented on the positive impact of implementation of MET or a RRS on the rate of unexpected cardiac arrests.14 18 22 There are several reasons that make it challenging to demonstrate a clear benefit in mortality associated with implementation of MET. The identification, triage and treatments of ‘at-risk’ hospitalised patients are complex and multi-factorial. The implementation of a MET may only represent one component of a larger hospital-led strategy for quality improvement and there may be unique contextual aspects of a given implementation in a given institution that is not measurable or generalisable. Likewise, it is challenging to capture the potential interaction and scope of involvement of MET with end-of-life care and its implications on the observed effectiveness in terms of in-hospital mortality.11 Lastly and similarly, capture of data on the ‘human’ aspects of front-line ward staff, in terms of beliefs about the MET and behaviours, that is recognition, interpretation and actions about the care of ‘at-risk’ patients, is also challenging and may inadvertently contribute to the introduction of bias and negatively impact generalisability.
Interpretation and clinical relevance
There are significant challenges when performing a study assessing the effectiveness of a complex intervention such as a MET. The MET involves coordination of organisational and logistical support. It also mandates broad acceptance by hospital staff and a significant ‘cultural shift’ in the management of ‘at-risk’ ward patients. Previous studies have shown that analyses of MET performance relatively early following MET implementation, such as in our study, may be flawed and non-representative of later performance.10 23 Acceptance by local ‘hospital culture’ will likely impact on the performance of MET over time.8 While there may be a ‘maturation process’ of the IL-MET model over time and there may be fewer interruptions of patient care in the ICU due to the MET intensivist attending to urgent consults on the ward, we were unable to unequivocally show this. There continues to be an ongoing need for systematic high-quality data capture on MET activity for quality assurance. Institutions should undergo regular re-evaluation of the effectiveness of the operational aspects and outcomes associated with the RRS or MET because each institution likely must act as its ‘own control’ given its unique environment, and because external validation and generalisability may not be possible.
Conclusion
In our single-centre experience, implementation of an IL-MET did not appear to significantly reduce the rate of in-hospital mortality or lengths of stay.
References
Footnotes
Funding Dr Bagshaw is supported by a Canada Research Chair in Critical Care Nephrology and Clinical Investigator Award from Alberta Innovates—Health Solutions (formerly Alberta Heritage Foundation for Medical Research).
Competing interests None.
Ethics approval Ethics approval was provided by University of Alberta Health Research Ethics Board.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data available on request from the corresponding author.