Analgesia & Resuscitation : Current ResearchISSN: 2324-903X

All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.

Research Article, Analg Resusc Curr Res S Vol: 2 Issue: 0

The Impact of Continuous Patient Monitoring at Various Times of Day on In-hospital Cardiac Arrest Mortality

Michael Mayette1 and Geoffrey K. Lighthall2,3*
1Department of Pulmonary and Critical Care Medicine, Stanford University School of Medicine, Stanford, CA 94305, USA
2Department of Anesthesia, Stanford University School of Medicine, Stanford, CA 94305, USA
3Department of Anesthesia and Perioperative Care, Veterans Affairs Medical Center, Palo Alto, CA 94304, USA
Corresponding author : Geoffrey K. Lighthall
Department of Anesthesia, MC 112A, Veterans Affairs Medical Center, 3801 Miranda Avenue, Palo Alto, CA 94304, USA
Tel: +650 493 5000x66756; Fax: +650 852 3423
E-mail: [email protected]
Received: May 20, 2013 Accepted: June 17, 2013 Published: June 24, 2013
Citation: Mayette M, Lighthall GK (2013) The Impact of Continuous Patient Monitoring at Various Times of Day on In-hospital Cardiac Arrest Mortality. Analg Resusc: Curr Res S1. doi:10.4172/2324-903X.S1-004

Abstract

The Impact of Continuous Patient Monitoring at Various Times of Day on In-hospital Cardiac Arrest Mortality

It is estimated that around 200000 treated in-hospital cardiac arrests occur annually in the United States, and this incidence may be increasing. Despite advances in management, unfavorable neurological outcomes and mortality remain high. Overall survival to hospital discharge is highly variable between studies, ranging from 0% to 42% with an estimated average in larger studies around 20% . Prior studies have demonstrated the association between criteria-defined abnormal vital signs on continuous monitoring and the incidence of cardiac arrest. The clinical benefit of monitoring on arrest outcomes was studied in a single-center study, with rates of survival to discharge in monitored wards arrests of 43.2% vs. 31.1% for non-monitored wards (p = 0.004). 

Keywords: In-hospital cardiac arrest; Continuous monitoring; Initial rhythm analysis; Time of day; Long-term outcomes; Clinical monitoring

Keywords

In-hospital cardiac arrest; Continuous monitoring; Initial rhythm analysis; Time of day; Long-term outcomes; Clinical monitoring

Introduction

It is estimated that around 200000 treated in-hospital cardiac arrests occur annually in the United States, and this incidence may be increasing [1]. Despite advances in management, unfavorable neurological outcomes and mortality remain high [2]. Overall survival to hospital discharge is highly variable between studies, ranging from 0% to 42% with an estimated average in larger studies around 20% [3-7]. Prior studies have demonstrated the association between criteria-defined abnormal vital signs on continuous monitoring and the incidence of cardiac arrest [8]. The clinical benefit of monitoring on arrest outcomes was studied in a single-center study [9], with rates of survival to discharge in monitored wards arrests of 43.2% vs. 31.1% for non-monitored wards (p = 0.004). The importance of continuous monitoring for at-risk patients was acknowledged in a recent consensus statement [10]. Other patient-specific (age, comorbidities such as sepsis, renal failure, cancer [11,12]) and reanimation-specific variables (time to response team arrival, initial shockable rhythm [5,13], location of arrest [12,14]) have been associated with better outcomes.
Time of the day also carries a major importance in cardiac arrests. First, incidence of arrests appears to increase in the early morning [15], possibly from diurnal fluctuations in neurohumoral substances such as catecholamines and cortisol. Outcomes of reanimation also appear to vary during the day, the worse prognosis being associated with nighttime arrests [16,17]. Multiple factors might explain this, including less house staff, inexperienced providers and longer delays between arrest and recognition.
This study presents the results of retrospective cohort analysis of the prior six years of in-hospital cardiac arrests to evaluate the impact of continuous monitoring, time of day, and initial rhythm on short-term and one-year mortality. We strove to better establish the complex relationships between these factors in the genesis and prognosis of cardiac arrests.

Methods

Setting
We conducted a retrospective analysis of a six-year period (October 2005 to January 30, 2011) of all non-ICU cardiac arrests. The center is a tertiary care teaching hospital within the US Veterans’ Affairs Health Care System with 325 beds, 15 ICU beds, 15 intermediate-ICU care beds, and an Emergency department that handles all case types except pediatrics and trauma. Continuous telemetery and oximetery are standard monitors in the Intermediate ICU, and approximately 15% of regular ward beds have the capability of continuous ECG monitoring. Data extracted from arrest records included date, time and location of arrest, initial rhythm identified, whether the arrest was witnessed, and whether the patient arrested while receiving any form of continuous monitoring. Death statistics were extracted from a mortality database that is updated monthly. The study received approval by the VA R&D committee.
Population
Comprehensive arrest records containing individual arrest data as well as date/ time / patient ID information kept by the hospital operators were developed in October 2006, the start of the study period. Ambiguous data was verified using the VA CPRS electronic medical records system. Arrest patients were purely an adult population (>18 years old). Hospitalized patients on any ward were included, as well as arrests in the emergency department as long as pre-arrest data were available (excluded if presented with out-ofhospital arrest).
All arrests for which there was reliable data on date and time, patient identification, location, and type of rhythm (up to the time of analysis) were assessed for inclusion. This data encompassed a period from October 2005 through September 30, 2011. Data was not analyzed if the “code blue” call was the result of false alarms, emergency situations without arrest, and cardiac arrests in DNR patients for whom no reanimation was performed. In cases of duplicate codes, only the most recent was considered in analysis of survival. All codes were considered in non-survival metrics such as location, rhythm type, and return of spontaneous circulation.
Statistical analysis
For analysis, presenting rhythms were split into categories of shockable rhythms (non-pulsatile ventricular tachycardia and fibrillation), non-shockable rhythms (pulseless electrical activity, symptomatic bradycardia, and asystole), and respiratory arrests requiring immediate intubation. Time of day was analyzed according to 8 hour time periods for further analysis (night is 00:00 hrs, day is 08:00-16:00 hrs, and evening is 16:00-24:00). All categorical variables were analyzed by Fisher’s exact test on a microcomputer using GraphPad Prism 5.0d software (GraphPad, San Diego, CA). Odds ratios were computed with Chi-square (χ2) test. Multivariate analyses, logistic regression model, were computed using SPSS 18.0 Software Package.

Results

From October 2005 through September 30 2011, 606 files for which a “code blue” was called were analyzed for inclusion. After excluding false codes, emergency situations without arrest, and cardiac arrests in DNR patients for whom no reanimation was performed, 124 arrest episodes were analyzed further. In this retrospective series, only 21 (16.9%) were initially shockable rhythms. Detailed breakdown of presentation rhythms as well as other metrics are shown in Table 1. Consistent with previous reports, fewer total arrests occurred during the evening period (16:00 to 24:00), as compared to night (00:00 to 08:00), and day (08:00 to 16:00) (15, 17). Events were evenly distributed across the week. The type of rhythm underlying arrests varied according to time of day. Shockable rhythms and respiratory arrests were 3-4 times more likely to occur between 08:00 and 16:00 than at other time intervals, as shown in Figure 1. The difference in the distribution of non-shockable vs. shockable rhythms between night and day arrests was statistically significant (p = 0.021). The pattern emerging from this data is that in the early morning hours, arresting patients are more likely to be discovered with a non-shockable rhythm (pulseless electrical activity, asystole or severe bradycardia) than with a shockable rhythm or respiratory decompensation.
Figure 1: Relative frequency of events (initial rhythm) normalized to time of day.
Table 1: Various metrics and demographics in arrest presentations (n = 124).
The composite 30-day mortality was 62.7 %, while overall mortality at 1 year after arrest was 73.7%. As shown in Table 2, mortality varied significantly according to the type of arrest. When compared with PEA and asystole, shockable rhythms have a significantly favorable 30-day mortality (odds ratio (OR) = 0.27, 95% CI 0.10-0.75, p = 0.0145). Respiratory arrests were found to have a similarly favorable 30d survival over non-shockable arrests (OR of 0.13, 95% CI 0.04-0.40, p = 0.0003). As shown in Figure 2, the time of arrest has a significant association with mortality. Nighttime arrests are more than twice as likely to result in mortality at 30 days when compared to daytime arrests (OR=5.12, 95% CI 1.85-14.14, p = 0.001). The 30-day mortality difference of nighttime arrests versus daytime remained statistically significant after adjustment for initial rhythm identified (shockable vs. non-shockable vs. respiratory) and for continuous monitoring (adjusted OR=4.44, 95% CI 1.49-13.23, p = 0.007).
Figure 2: Kaplan-Meier estimates of cumulative 1 year mortality according to time of day of the arrst.
Table 2: 30-day mortality according to arrest scenario (rhythm, time) (n = 118).
Kaplan-Meier estimates of cumulative mortality (Figure 2) show that the difference in mortality of night versus daytime arrests persists to at least one year after the event (p = 0.002). Evening arrests have similar mortality when compared to day (p = 0.25). Mortality at either 24 hours, 30 days or 1-year did not appear to be impacted by whether the arrest occurred on a weekday versus weekend.
Table 3 demonstrates the effect of continuous monitoring, either cardiac telemetry or oximetry, on mortality in this cohort. No significant effect can be identified on the group of patients as a whole, nor can it be demonstrated on a specific stratum of patients when separated according to time of day or initial rhythm. A non-significant trend suggests an association between monitored shockable rhythms and improved survival, but overall our numbers are too small to draw any conclusions (RR for monitored patients 0.43, 95% CI 0.17-1.10).
Table 3: Mortality in monitored and not monitored patients.
A multivariate logistic regression model was created to identify variables independently associated with higher mortality. Only nighttime arrests and arrests presenting with non-shockable rhythms were independently associated with mortality, as shown in Table 4.
Table 4: Logistic regression multivariate analysis of independent risk factors for 30-day mortality (n = 118).

Discussion

This study was aimed to study the epidemiology and predictive factors associated with short and long-term mortality of in-hospital cardiac arrests. Our work highlights existing studies showing a significantly higher mortality for arrests presenting as non-shockable rhythms, or occurring at night [5,17]. Unlike previous studies on inhospital arrests, our patients were followed up to one year following the arrest. We were thereby able to demonstrate that the magnitude of increased mortality after nighttime arrest, up to one year after index event, was much higher in our study than in previously reported literature (OR for day vs. night mortality was 5.12 (95% CI 1.85- 14.14) in our study vs. 1.43 (1.38-1.49 95% CI) in Peberdy et al. [17]).
As previously reported, our study found that arrests in which the first rhythm identified is shockable (ventricular fibrillation or pulseless tachycardia) are much more prevalent during the day [15,17]. Explanations of this finding range from the biologic (different pathophysiology for arrests occurring at night versus day), or systemic factors such as the possibility that shockable rhythms or respiratory insufficiency occurring at night goes unrecognized for longer periods of time before treatment is instituted, with many “evolving” to nonshockable rhythms carrying a dimmer prognosis. In our study, the difference in mortality persisted after adjustment for initial rhythm and monitoring, suggesting global management of patients might be different during the night, leading to higher mortality. Such differences could include less experienced personnel at night, higher performance of arrest teams during regular working hours, or more difficult access to advanced techniques during the night.
Multivariate regression analysis identified only two variables independently associated with higher mortality, namely nighttime arrests and initially non-shockable rhythms. There is strong trend although not statistically significant, suggesting that respiratory arrests behave like shockable rhythms in having a lower associated mortality.
The benefit of continuous monitoring such as ECG or pulse oximetery for hospitalized patients appears intuitively obvious. Based on expert agreement, it has been recommended that such monitoring should only be limited by practicality and affordability [10]. However, apart from invasive monitoring in the ICU, evidence-based demonstration of benefit from such practice on mortality remains elusive. We tried to identify a benefit of continuous monitoring on survival in our cohort, however mortality in previously monitored and not-monitored patients were statistically similar, and multivariate analysis could not demonstrate an independent benefit of monitoring on 30-day or one-year mortality. Larger studies are needed to firmly establish or refute the impact of continuous monitoring on arrest survival. Needless to say, analysis of end points such as organ failure, length of stay and other morbidities may more appropriately address the value of increased monitoring.
Long-term follow-up of patient survival was thorough. From survival curves, we can see that the vast majority of mortality occurred very close to the time of the arrest. If a patient survived the initial days following the arrest, 30-day mortality was strongly correlated to 1-year mortality (overall mortality at 30 days was 62.7 %, versus 73.7 % at 1 year). Previous studies demonstrated a steady decline in survival after arrests, with recurrent cardiac arrests as the main cause [18,19]. Our study supports the importance of post-arrest care, follow-up, treatment of comorbidities in high-risk patients, as well as prevention of recurrence.
This study illustrates the difficulty in studying in-hospital cardiac arrests. They evolve from various different clinical scenarios, which all carry different prognoses, as well as various underlying pathologies which impact outcomes and survival. Our study has the benefit of studying a rather homogeneous population (the vast majority being older males) and for a longer period of time than prior studies.
This study is certainly limited in power due to small number of patients, especially when study groups were subdivided according to arrest types and monitoring status. Also, long-term effects (up to 1 year) should be interpreted with caution, since many other health factors, unknown and not accounted for, might have intervened in the outcomes.

Conclusion

In-hospital cardiac arrests continue to carry a high mortality rate, both short and long-term, despite advances in reanimation. Once initial in-hospital mortality has passed, long-term follow-up shows small but steady added mortality, suggesting improvements in postarrest care can be made. Time of the day was found to be an important factor associated with the presenting rhythm and associated mortality. Arrests occurring during the day are more than twice as likely to have a favorable outcome. The benefits of continuous monitoring remain elusive.

References




















Track Your Manuscript