| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Circulation. 1997;95:2213-2239.)
© 1997 American Heart Association, Inc.
Articles |
Recommended Guidelines for Reviewing, Reporting, and Conducting Research on In-Hospital Resuscitation: The In-Hospital `Utstein Style'
A Statement for Healthcare Professionals From the American Heart Association, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, the Australian Resuscitation Council, and the Resuscitation Councils of Southern Africa
Key Words: defibrillation • AHA Medical/Scientific Statements • cardiopulmonary resuscitation • sudden death
 |
Introduction |
|---|
Top
Introduction
BackgroundThis scientific statement is the product of the Utstein '95 Symposium held June 23-24, 1995, at Utstein Abbey, Island of Mosteroy, Rogaland County, Norway. Draft versions were circulated for comment to participants of the Utstein '95 Symposium; the European Resuscitation Council Executive Committee; the Emergency Cardiac Care Committee of the American Heart Association; the Executive Committees of the Heart and Stroke Foundation of Canada, the Australian Resuscitation Council, and the Resuscitation Councils of Southern Africa; and several outside reviewers. The development of this statement was authorized by the Science Advisory and Coordinating Committee of the AHA and the Executive Committee of the European Resuscitation Council.
We do not know the true effectiveness of in-hospital resuscitation. Observed results of the many published studies vary greatly. Studies originate from different settings and have different patient populations. Reports suffer from nonuniform nomenclature and variable inclusion definitions. Patients differ in the extent of comorbid conditions and interventions in place at the time of cardiac arrest. These differences prevent valid interhospital and intrahospital comparisons and make determining the effectiveness of current resuscitation techniques impossible.
To develop these guidelines the task force used a consensus development process that originated with the "Utstein style" for reporting outcome data from out-of-hospital resuscitation events. Task force members performed an integrated review of published studies. An initial draft was prepared, discussed, and revised at a 2-day conference. Further drafts were revised and circulated among task force members and discussed face-to-face at three subsequent meetings.
The task force defined a set of data elements that are essential or desirable for documenting in-hospital cardiac arrest. Data categories are hospital variables, patient variables, arrest variables, and outcome variables. The "In-Hospital Utstein-Style Template" was developed to summarize these data and recommendations for reporting a specific set of survival rates and outcomes. The task force recommends that four critical intervals be included, whenever possible, in all reports of in-hospital resuscitation: collapse to cardiopulmonary resuscitation (CPR), collapse to first defibrillation, collapse to advanced airway management, and collapse to administration of first resuscitation medications. The task force also developed the Utstein-style "Standard Reporting of In-Hospital Cardiopulmonary Resuscitation" form, which hospitals can use to record cardiac arrest data for individual patients.
Collection of the recommended variables will enable intrahospital and interhospital comparisons and support national and international research. Collection and review of these data should improve patient care and event documentation for individual patients and reduce medical-legal risks. These recommendations, however, are only a beginning. Definitive recommendations for in-hospital resuscitation require additional research. CPR remains a dramatic medical intervention with both indications and contraindications. Resuscitation activities must be conducted ethically, with concern for the principles of patient self-determination and death with dignity and for the sensitivities of family and friends.
|
Background |
|---|
Growth of Resuscitation Activities
Since the rediscovery of the effectiveness of closed chest compression in 1960, medical personnel have made CPR one of the most frequently performed medical interventions in the world. Organizations such as the AHA, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, the Australian Resuscitation Council, and the Resuscitation Councils of Southern Africa have developed practical clinical guidelines for emergency cardiac care.
The Emergency Cardiac Care programs of the AHA and equivalent training programs teach techniques of basic life support (BLS) and advanced life support (ALS) to both in-hospital and out-of-hospital emergency personnel. These programs serve to encourage citizen CPR. Versions of these programs have spread to most communities and hospitals in the developed world. Training equipment, educational materials, and personnel costs associated with this training represent more than a billion dollar "industry" in the United States alone.1 The resuscitation councils of many countries have implemented similar programs and are establishing training networks to ensure that hospital professionals learn and can use recommended procedures.
In the United States hospital accreditation requirements specifically mandate that hospitals develop a planned response to in-hospital cardiac emergencies. Most accredited US hospitals require that medical and nursing staff who respond to cardiac arrests be trained in BLS and advanced cardiac life support (ACLS). This requirement has encouraged the growth of a large emergency cardiac care training industry in the United States. Many hospitals in Europe, Australia, Canada, and South Africa, as well as in countries outside these areas, have mandated or plan to establish similar programs.
In addition, resuscitation is an active and growing subject of academic inquiry. Throughout the world researchers seek the most effective interventions and techniques to resuscitate people after cardiopulmonary emergencies. The journal Resuscitation publishes reports of such research, and other major medical journals devote considerable attention to resuscitation topics.
Nevertheless, the true effectiveness of many aspects of resuscitation remains unknown despite the huge investment of scientific and healthcare resources.2 3 4 5 6 7 8 9 For example, we do not know the value of antiarrhythmic or adrenergic agents given for cardiac arrest,9 10 11 nor do we know the true effectiveness of training programs in emergency cardiac care.3 12 How well do people learn, and can they remember what they learn?5 Will they perform these skills in a true emergency?4 13 14 Although several investigators have observed that the quality of resuscitation attempts in terms of objective process criteria improve after formal CPR training,6 7 8 significant differences in the critical outcome of survival rates have not been recorded. Finally, it is not firmly established that the correct use of resuscitation protocols, even by formally trained emergency personnel, influences patient outcome.2
During the late 1980s and early 1990s, resuscitation researchers began to systematically examine the effectiveness of current resuscitation procedures.15 The International Liaison Committee on Resuscitation has sought to evaluate available scientific data related to resuscitation protocols.16 A continuing scientific review of animal and human resuscitation research has led investigators to ask whether the data justify changes or additions to the clinical guidelines. This effort is an attempt to remove anecdotal evidence, subjective impressions, and proclaimed affirmations from resuscitation protocols.
This critical review has been sobering. Few resuscitation interventions are based on valid scientific data. Proper randomized, controlled clinical trials focused on sudden cardiac death, and unexpected resuscitations are difficult, if not impossible, to perform. Many existing resuscitation practices are driven by pathophysiological and nonquantitative reasoning rather than by evidence-based medicine. Resuscitation guidelines often present "best guesses" rather than authoritative conclusions.
Results From In-Hospital Resuscitation Efforts: Variations in
Reported Success
For more than 30 years researchers have published many studies on
survival after in-hospital CPR. Until recently no clear picture of
success had emerged. Three major reviews of more than 50 published
articles on survival after in-hospital CPR have demonstrated wide
variations in survival.17 18 19 McGrath17
calculated survival rates of 38% at 24 hours (range 13% to 59%) and
15% at hospital discharge (range 3% to 27%). DeBard18
reported survival rates of 39% at 24 hours and 17% at discharge to
home. Cummins and Graves19 reviewed 44 studies and
calculated survival rates to hospital discharge that ranged from 3% to
27% following an in-hospital cardiac arrest. Such wide variations in
the rate of survival are explained largely by marked differences in
inclusion criteria and outcome definitions.
Two large-scale projects provide the best evidence on success of in-hospital resuscitation. The Belgian Cardiopulmonary Cerebral Resuscitation Registry (BCCRR), working with the European Resuscitation Council, gathers information on survival from both in-hospital and out-of-hospital resuscitation.20 21 In the British Hospital Resuscitation Study (BRESUS), investigators analyzed the results of 3765 attempted cardiopulmonary resuscitations in 12 teaching and nonteaching hospitals throughout the United Kingdom.22 Neither of these projects, however, is exclusively focused on in-hospital resuscitation. In BRESUS 25% of patients experienced onset of arrest during the prehospital phase. Investigators observed that 39% of patients survived the immediate arrest, 28% were alive 24 hours later, 17% were discharged from the hospital, and 12.5% survived for 1 year. Both BRESUS and BCCRR helped pioneer standard methods of recording arrests for audit, clinical trials, and community studies. These studies provide a yardstick for comparing results from selected UK and Belgian hospitals with those obtained in other countries and for helping assess the effect of subsequent management changes.
Origin of the Utstein Style and Extension to In-Hospital
Arrests
The Utstein style for reporting cardiac arrests arose from a 1990
conference at the ancient abbey of that name on an island near
Stavanger, Norway. That conference and another later that year were
attended by representatives of the AHA, the European
Resuscitation Council, the Heart and Stroke Foundation of Canada, and
the Australian Resuscitation Council. The results of these meetings
have been previously reported.23 24 25 26 27 28 29 30 The major concern was
that the results of resuscitation endeavors in different countries, and
even within countries, could not be compared meaningfully. Researchers
have used disparate end points to assess the effectiveness of different
systems and interventions. Useful comparisons have been prevented by
this lack of uniform definitions and standard methodologies.
The Utstein style has attracted wide interest and has become a familiar term among members of the resuscitation community. Many researchers and system directors have adopted the Utstein templates, style, and nomenclature to report results of prehospital resuscitation. The success of this international initiative soon led to uniform international styles for reporting the results of pediatric resuscitation31 and experimental (laboratory) resuscitation.32
This report continues the Utstein style process by including adult in-hospital resuscitation within the international reporting agreements. These guidelines and recommendations can improve the scientific design of research projects and increase the clinical usefulness of published studies.33 Improved projects and publications will provide consistent and reliable evidence on which to base treatment decisions. Uniform definitions and methodology will support valid interpretations of findings across multiple studies and permit more accurate integrative reviews and meta-analyses of resuscitation studies.
In-Hospital Resuscitation
In-hospital resuscitation presents unique challenges for
research and evaluation. In 1990 the original Utstein Task Force
considered including in-hospital resuscitation in the uniform
guidelines24 but postponed doing so because of its
daunting complexity. Critical problems needed to be resolved:
inconsistent definitions used in the description of
resuscitation emergencies and outcomes, comorbid factors and severity
adjustments for the victims, and the effects of interventions already
in place at the time of arrest.
A cardiac arrest is defined in the Utstein style as "the cessation of cardiac mechanical activity . . . . confirmed by the absence of a detectable pulse, unresponsiveness, and apnea (or agonal respirations)."24 30 A hospital patient, however, may have many degrees of cardiac and pulmonary dysfunction that cannot be characterized as cardiac arrest, such as hypotension or shock. A patient's respiratory status can range from normal to agonal or gasping breaths to apnea, or the patient may be on artificial ventilation.
A hospital patient needing resuscitation could have comorbid conditions that may or may not influence resuscitation outcome. For example, comorbid conditions could be unrelated to resuscitation outcome (eg, hysterectomy for uterine fibromas), moderately related (eg, acute pneumococcal pneumonia), or strongly related (eg, acute pulmonary edema after myocardial infarction). Moreover, comorbid and preexisting conditions may vary greatly in severity; for example, acute pneumococcal pneumonia may vary greatly in extent of organ involvement and in severity. Almost every publication on in-hospital resuscitation has neglected to allow for the magnitude of comorbid conditions.
|
In-Hospital Utstein Variables |
|---|
To improve reports of in-hospital resuscitation, the Utstein Task Force identified four sets of variables that required further definition: hospital variables, patient variables, arrest variables, and outcome variables (Fig 1
).
|
No hospital, researcher, quality improvement advocate, or process evaluator will be able to collect all of these variables. Much of the data discussed will be collected only for specific purposes. For example, resuscitation attempts are critical events in the care of a patient that should be recorded as a vital part of the patient's medical record. Often these data are recorded incompletely and inaccurately in the rushed turmoil of in-hospital resuscitation. The minimum data elements related to individual resuscitation attempts that should be collected are defined.
Outcome measurements for interhospital or intrahospital comparisons or specific research projects require a description of the context of the resuscitation. This context includes hospital variables as well as patient variables. Definitive evaluation of resuscitation efforts requires measurement of outcome variables, either immediate, intermediate, or long-term. Outcome evaluations, however, require following the patient outside the hospital, a task that requires a large amount of time, energy, and personnel resources.
Patient Variables
Table 1
lists the patient
variables that the task force recommends be reported. In selecting
the required variables, the task force sought to include
variables for which there is evidence of an association with
differences in survival rates. Table 1 defines these variables and
provides directions or comments for their use. Many of these
variables are self-explanatory and descriptive. The definitions and
directions for some variables are expanded below.
|
Age
Some studies have linked older age to poor outcome following
out-of-hospital cardiopulmonary arrest and
resuscitation.34 35 36 37 In other studies, however, age alone
was not shown to be an independent determinant of
survival.38 39 40 41 42 43 44 45 46 47 The observed differences in the effect of
age on out-of-hospital and in-hospital arrest survival are difficult to
separate from the effects of comorbid conditions.47
Age categories should, at a minimum, separate pediatric data from adult data and should stratify the adult population. Adolescents are a difficult group to classify.14 15 16 17 18 19 Studies on the incidence of ventricular fibrillation (VF) in various age groups, for example, have yielded conflicting results. These differing results stem, in some degree, from whether to group adolescents with adults or children.48 49 50 51 52 The context of the study determines the designation: adolescents may be included in pediatric studies provided they are separated from younger children or in adult studies provided they are separated from older adults. Because no single age is consistently used to separate children from adults, the following age categories and subsets (the convention is that the year of age advances on the exact anniversary of birth) are recommended:
- Infant: 0 to 12 months
- Report the following age subsets:
0 to 30 days
- 31 days to 12 months
- Child: 1 to <20 years
- Report the following age subsets:
- 1 to <3 years
- 3 to <8 years
- 8 to <14 years
- 14 to <20 years
- Adult: 20 years and older
- Report the following age subsets:
- 20 to <25 years
- 25 to <35 years
- 35 to <45 years
- 45 to <55 years
- 55 to <65 years
- 65 to <75 years
- 75 to <85 years
- 85 and older
Gender
Researchers have paid increasing attention to the role of gender
in cardiovascular disease in terms of risk
factors,53 54 outcomes,53 54 55 56 57 58 59 and treatment
differences.60 61 62 63 64 65 66 The task force recommends
recording the gender of all patients.
Witnessing and Monitoring of Event
Witnessed cardiac arrests are those that are seen, heard, or
monitored. All witnessed arrests should be further classified as
monitored (more often in intensive care units [ICUs] or telemetry
units) or unmonitored (more frequently in locations that involve less
critical care) because monitoring is likely to influence other patient
variables. Patients whose cardiac arrests are not witnessed have
markedly reduced chances of successful resuscitation and should be
considered separately.
Location of Event
In view of the differences in case mix and likely response
interval in different areas of the hospital, documentation of the
patient's location at the time of cardiac arrest is considered an
essential patient variable. Table 1
lists recommended location
categories.
Advanced Life Support Interventions in Place at Time of
Arrest
Some resuscitation interventions are often instituted before a
patient experiences an actual resuscitation emergency. For example, any
of the following interventions may be in place when resuscitation
begins: monitor and defibrillator, mechanical ventilation,
intravenous pressor or inotropic agents, or
intravenous antiarrhythmic agents. Survival rates for such
a heterogeneous group of patients will likely be affected
by the presence of these interventions as well as by the condition of
the patients before the emergency. To illustrate, the outcome
probability for VF arrest will likely differ markedly for a patient who
is intubated and on a ventilator compared with a patient without
special surveillance in a general ward. Table 1 describes a number of
interventions that should be noted if present at the time of
arrest.
Previous Cardiopulmonary Arrest
Survival of a recent cardiac arrest constitutes a significant
comorbid condition for future arrests and resuscitations.
Recommendations on how to analyze patients who have had
multiple arrests and have undergone previous resuscitations are
discussed in the section on template box 9.
Reason for Admission
The task force recommends placing patients into several categories
on the basis of the major reason for hospital admission (Table 1).
These categories provide information on significant comorbid conditions
that would otherwise confuse the results of resuscitation attempts. For
example, patients hospitalized for medical reasons tend to have
different causes of arrest and different resuscitation outcomes than
patients admitted for surgery or out-of-hospital trauma.
Prearrest Functional Capacity
Resuscitation outcomes can be evaluated only by comparing
postresuscitation status with prearrest status.67 A
simple, validated tool for this purpose is lacking. The formal tools
that can be used after the arrest are almost never applied before
arrest. Furthermore, the optimal timing of the prearrest evaluation of
function is unclear: should the level of function be recorded
immediately before the arrest, at the time of admission, or at some
time before admission?
The Glasgow-Pittsburgh outcome categories, including the Glasgow Coma Score,68 69 70 the Cerebral Performance Categories (CPCs), and the Overall Performance Categories (OPCs)71 72 are the most widely used tools to evaluate functional outcome following resuscitation in adults. Numerous out-of-hospital studies have followed the recommendations of the Utstein style in this respect.73 These categories were modified and validated for use in children.74 They appear sufficiently broad so that the required information may be reliably derived from retrospective chart review, although this has not been confirmed. The task force recommends some attempt to record, through chart review or patient or family interview, the CPC and the OPC (using pediatric modifications if appropriate) at the time of hospital admission.
Comorbid Conditions
Comorbid conditions have a powerful influence on the outcome of
patients treated for out-of-hospital VF.75 A comorbidity
index derived from a medical history of 10 chronic conditions,
medication use, and recent symptoms was strongly linked to
survival.75 Only one published study of in-hospital arrest
patients considered these comorbid conditions.47
Nevertheless, the task force recommends recording the presence
of major comorbid conditions. The likelihood that these comorbid
conditions will strongly influence resuscitation outcomes possesses a
high degree of face validity. No single validated method exists to
describe comorbidity in hospitalized patients. Although this is an
important area for future research, the issue of comorbidity cannot be
ignored while awaiting future studies. The task force recommends using
the International Classification of Diseases, ninth edition (ICD-9-CM),
to describe comorbid conditions. Because ICD-9-CM coding is completed
only at discharge or death of the patient, care must be taken to
include only the major ICD-9-CM codes present before the
arrest.
The ICD-9-CM codes, however, have limitations. In many hospitals ICD-9-CM codes and discharge and diagnosis data (often collected for insurance purposes) are generated more to calculate hospital charges than to document patient diagnoses.76 77 These codes consist of a heterogeneous group of conditions, including diagnoses, pathological processes, symptoms, physical findings, test findings, and severity indicators. The "correct" coding may vary among providers, institutions, or regions. Researchers need some method to group ICD-9-CM codes to enable meaningful comparisons of outcomes between population subgroups. The BRESUS investigators used eight groupings of ICD-9-CM codes to classify a large number of ICD-9-CM codes.22 This grouping may be useful, although others may be valid.
Severity Estimates
The severity of a patient's prearrest condition may certainly
affect clinical outcome. No simple, validated severity classification
system is appropriate for both ICU and non-ICU patients. The Utstein
Task Force considered a number of existing severity classification
systems. The Acute Physiology and Chronic Health Evaluation (APACHE II
or III) scoring system for patients in the ICU78 79 80 is
used in many critical care units, but it is complex and designed solely
for ICU patients. Some research has addressed the problem of predicting
failure to survive after in-hospital CPR.41 81 82 83 84 The
Pre-Arrest Morbidity (PAM) Score includes comorbidity factors but has
been validated only as an instrument to predict outcome following
cardiac arrest.83 85 The Prognosis-After-Resuscitation
(PAR) Score has fewer variables than the PAM score and in Ireland
was better at predicting patients unable to survive in-hospital
resuscitation.84 The Therapeutic Intervention Scoring
System (TISS)86 87 and the modified TISS (for non-ICU
patients) are limited because they reflect more the therapy provided
than the severity of the patient's illness. The various revisions of
the New York Heart Association functional classification are too
nonspecific and broad.88 At this time the task force
recommends use of the most appropriate validated severity
classification system for the population studied.
Other arrest data, including location (ICU versus ward) and concurrent therapeutic interventions (eg, mechanical ventilation or administration of vasopressors) (see below) may provide some surrogate information to suggest the severity of illness. Research is necessary to establish whether these proxy measures are valid surrogates for classification of severity.
Race/Ethnicity
Several research projects have identified a powerful
association between race and the incidence of bystander
CPR,89 outcomes of cardiovascular
disease,90 91 92 survival of out-of-hospital sudden cardiac
arrest,93 94 and survival following in-hospital
CPR.95 The outcome of resuscitation attempts for some
ethnic groups may be influenced by additional factors such as
socioeconomic status,96 97 inequalities in access to
health care, utilization of health care, or provision of health
care.
The task force recognized major differences in the categories for expressing race among countries. The racial mix of patients in different countries varies significantly, and no common list of racial categories that applies to all countries can be specified. Any group accounting for only a small proportion of a study population will seldom produce data that can be analyzed meaningfully. The task force therefore recommends that researchers specify only those minorities or ethnic groups that make up a significant proportion of the total population of a study.
Socioeconomic Status
Some researchers have found an association between lower
socioeconomic status and lower rates of survival of out-of-hospital
cardiac arrest.97 98 Researchers have used a variety of
measures of socioeconomic class in other epidemiological research, none
of which are internationally accepted. The task force does not
recommend a single measure of socioeconomic status to collect but
considers this an interesting area for future research.
Event Variables
The major event variables that should be recorded are
rhythms, interventions performed, event times, and event intervals.
Rhythms
The task force recommends grouping arrest rhythms into two major
categories: VF/pulseless ventricular
tachycardia (VF/VT) or non-VF/VT.15 16
Non-VF/VT can be subclassified as either pulseless electrical activity
(PEA) or asystole. These rhythms are defined in template boxes 5 and
6.
Interventions
Essential interventions that should always be recorded if
performed include the following:
- Defibrillation
- Insertion of an artificial airway, specified by type (eg, endotracheal tube or laryngeal mask)
- Drugs, including antiarrhythmic agents and vasopressors
- Pacing, specified by type (eg, external or internal)
- Mechanical circulatory assist devices (eg, balloon pump or cardiopulmonary bypass).
Event Times and Event Intervals
Table 1 provides definitions and directions for the recommended
variables. All of the following event times are essential and
should be recorded if relevant and available:
- Time of collapse (event onset)
- Time of call for help
- Time of arrival of CPR team
- Time of confirmed cardiac arrest
- Time CPR started
- Time CPR stopped
- Time of first defibrillatory shock (and all subsequent shocks)
- Time of advanced airway management (including intubation)
- Time of first dose of intravenous epinephrine/adrenaline and all other resuscitation medications
- Time of any return of spontaneous circulation (ROSC) and of sustained ROSC
- Time of end of ROSC
- Time of abandonment of resuscitation attempt
Lack of clock accuracy and synchronization can be a major problem in resuscitation research.99 100 101 Hospitals should synchronize the clocks used in resuscitation records to a single time source. Current defibrillator models provide time annotations on monitor strips; therefore, the defibrillator is the best source for clock synchronization.
From the recorded event times a variety of intervals (time elapsed between two events) can be calculated. The following intervals are recommended as the gold standard process variables for interhospital and intrahospital comparisons:
- Interval from event onset to start of CPR
- Interval from event onset to first defibrillation
- Interval from event onset to advanced airway management
- Interval from event onset to first administration of resuscitation medication
Other clinically important intervals include the following:
- Interval from event onset to ROSC (palpable spontaneous pulse)
- Interval of sustained ROSC
- Interval from event onset to CPR stopped
Outcome Variables
Resuscitation aims to return patients to their level of health
before the cardiac arrest. Resuscitation outcome can be expressed in
one or more of three domains: survival (did the patient live?),
longevity (how long did the patient live?),70 102 103 and
quality of life (how well did the patient live?).104 105 106
Survival can be subdivided further using a time dimension, such as
immediate (ROSC), short-term (discharged alive from the hospital), or
long term (6 to 12 months).
Quality of Life as an Outcome
Quality-of-life outcomes reflect the widely accepted concept that
health is not just the absence of disease and infirmity but also the
presence of physical, mental, and social well-being.107
Assessment of quality of life as an outcome requires reference to
physical, psychological, and social domains of health. These domains
are measured as either objective assessments of functioning or as
subjective perceptions. The rich interaction of objective functioning
and subjective perceptions determine the overall quality of life for an
individual. Two people with the same results on objective measures of
health may have profoundly different estimates of quality of life.
Functional Outcomes
Although many valid functional assessment instruments have been
developed, no single approach applies to all patients. The Glasgow Coma
Score, introduced and validated for neurological assessment following
trauma, is a simple-to-record, reproducible assessment of
consciousness.70 The three individual components of the
15-point score (motor response, verbal response, and eye opening) are
recorded separately, then in total, 24 hours after the initial
event. The motor component is best associated with outcome,
particularly if the score is recorded daily over
time.108 109 Recording of brain stem reflexes may
add additional predictive value.110
The CPC score is one of several instruments developed to assess cerebral performance outcomes following both traumatic and anoxic cerebral injuries.71 72 This score approaches a disease-specific instrument because survival of cardiac arrest constitutes survival of a transient, global cerebral ischemic or anoxic event. The CPC score has been used to evaluate outcomes in several studies of cardiac arrest survivors.105 111 112 113 114 115 However, its use as a single measure of neurological outcomes in cardiac arrest survivors has been strongly criticized.116 Hsu et al116 noted that in a small group (n=35) of cardiac arrest survivors, the CPC score correlated poorly with subjective quality-of-life assessments and results of validated objective functional testing instruments. An additional useful measure of minimum neurological outcome is the time to awakening or return of consciousness.117
As minimum uniform assessments, the task force recommends recording the time to awakening and the serial Glasgow Coma Scores in the immediate postarrest period. These measures will document either return of consciousness or persistence of various levels of the coma state. In addition, the CPC should be recorded at hospital discharge and again 6 months to 1 year after the event. The Utstein Task Force recognizes the limitations of the CPC score and that use of multiple standardized testing instruments is superior.106 118 The CPC score has the appeal of simplicity and practicality.
For patients who die between hospital discharge and 1 year after the arrest, the best functional status during that year should be recorded if it is available and practical to obtain. Recording the best functional status is recommended because a subsequent decline in functional outcome after initial improvement may be unrelated to the arrest. For example, premorbid conditions may affect eventual outcomes more than the arrest episode itself.
Additional Outcome Assessments
Length of stay (days of hospitalization) after resuscitation is a
useful variable to record. Length of stay can help answer
questions about the value of in-hospital resuscitation because a number
of studies have documented prolonged hospital stays after
resuscitation.114 119 120 Patients may be resuscitated
from a cardiac arrest but condemned to live many weeks in a severely
compromised state, requiring costly intensive care. Serious
cost-benefit concerns arise when compromised patients die after
prolonged hospital stays, not to mention the pain and suffering
experienced by the patient and family members.
Cost data have not been included in many publications on resuscitation.114 121 122 The cost-effectiveness of interventions such as CPR and resuscitation, however, must be evaluated. Investigators should measure costs rather than charges because costs better represent the societal burden of an intervention such as CPR.123 The cost benefit of in-hospital resuscitation may be expressed as the cost per survivor to hospital discharge or the cost per quality-adjusted life-year. Calculation of total costs includes the costs of resuscitation, stay in the ICU, stay in the ward, and subsequent nursing home or home health costs. An increase in survival rates will result in lower costs per survivor. The value of this information may be severely limited by economic factors such as inflation, profitability issues, cost shifting, reimbursement policies, and variable national approaches to healthcare financing.
If the patient is discharged, researchers may record the discharge destination: home (or preevent residence), rehabilitation facility, extended-care facility (nursing home), another acute care hospital, or other. Although discharge destination is often used as a surrogate for neurological outcome, researchers should record the need for home nursing care because discharge to home may not necessarily represent a good outcome.
Hospital Variables
Hospital variables are required for interhospital comparisons
because the setting of an arrest influences resuscitation capabilities
and outcome data. However, no widely accepted categorization of
hospital variables exists. Many variables are independent
factors that may affect outcome. Table 2 provides a checklist of information to include on reports of
in-hospital resuscitation. The task force declined to recommend any of
these variables as essential because the effect of any of these on
arrest outcomes is speculative. Additional research and publications
will provide better guidance.
|
|
Utstein Template for Reporting Data From In-Hospital Resuscitation |
|---|
Template Concept: What It Represents, How to Use It
The Utstein style was created for reporting out-of-hospital cardiac arrest.24 30 The centerpiece of that work was a template for uniform display of outcomes. This template, although now used widely, is not always used correctly. The template concept is based on the following principles:
The initial boxes of the template define a
population; this population must be accounted for in subsequent boxes
of the template.
The number of each box serves both as the
numerator for the box above it and as the denominator for the box below
it.
The template displays a central
"trunk," which is considered the most important subpopulation to
report.
Depending on the objectives of the study,
several branches can be identified and analyzed, analogous to a
hierarchical filing system or pull-down menus in the organization of
computer files.
Although numerous branches of the algorithm
can be developed, in most instances this is not necessary.
Concerns about sample size will necessitate
combining rather than splitting population subsets. Investigators
should make clear which subsets are being considered and which have
been combined or subdivided.
Template Boxes (Fig 2)
Template Box 1: In-Hospital Patients With a Pulse
This template box provides the conceptual starting point to
analyze resuscitation results obtained in a hospital.
Researchers do not need to determine this exact number because it is
not required for later outcome calculations. This population is
analogous to the "population served" in out-of-hospital
statistics.24 30 Any patient arriving at the hospital with
a spontaneous pulse becomes a candidate for later resuscitation
attempts. Include all patients who occupy a hospital bed. No minimum
stay, such as more than 24 hours, is recommended. Include patients who
experience an arrest in the emergency department or casualty suite.
Exclude from analysis (or report as separate subsets if these patients represent an area of particular interest) persons brought to the hospital for pronouncement of death and persons who do not occupy a bed, such as outpatients or clinic attendees who suffer a cardiac arrest inside the hospital. Finally, report separately persons who originally had an arrest outside the hospital and were brought to the emergency department after ROSC or with continuing CPR. These patients more properly belong in the original out-of-hospital template.
Template Box 2: Cardiac Arrest, No Resuscitation Attempted
- Patients designated as DNAR
- Patients not designated as DNAR
- Found dead
- Considered futile
Some patients, for reasons of self-determination, living wills, advance directives, or medical indications, should not receive attempted resuscitation. Most hospitals have mechanisms to identify these patients. These patients do not belong in a denominator for calculating in-hospital survival rates and should be separated from the main branch of the template.
The task force endorses use of the term "do not attempt resuscitation" (DNAR). It is inappropriate and misleading to use a term such as "do not resuscitate," which implies to patients, family, and friends that personnel could have resuscitated a patient if they had wished. The term DNAR offers a more accurate assessment of the reality of resuscitation attempts.124
It is recognized that there will be additional patients for whom no resuscitation attempts will be made. The two major groups are patients found dead in bed or elsewhere in the hospital and patients for whom the response teams, for a variety of reasons, considered resuscitation efforts futile.
Template Box 3: Attempted In-Hospital Resuscitations
- Total
- Defibrillation only
- Chest compressions only
- Airway interventions only
- Combination interventions
Any meaningful resuscitation attempt by appropriate healthcare providers qualifies as attempted in-hospital resuscitation (template box 3). Stopping resuscitation efforts because the patient failed to respond to initial ACLS efforts represents an attempted resuscitation. If rescuers inadvertently provided a full resuscitation attempt to a DNAR patient, this is an attempt. If only a momentary attempt was made while personnel confirmed the DNAR status or because of immediate discovery or determination that resuscitation is not indicated, the "no resuscitation attempted" classification should be selected.
The task force discussed a single, easily recognized criterion to define a resuscitation attempt. Published work suggests that "chest compressions started" would provide an unambiguous criterion.125 In addition, chest compressions started is consistent with the out-of-hospital Utstein nomenclature for the denominator starting point.24 30 The disadvantage of the chest compression started criterion, however, is that it excludes many patients who do not require chest compressions because personnel resuscitate them by defibrillation or airway interventions only. For example, patients with monitored pulseless VT or VF may be treated successfully by immediate defibrillation. This action would obviate the need for chest compressions.
Respiratory compromise and deterioration, especially in children, could lead to bradycardia, hypotension, asystole, and full cardiac arrest.126 Timely airway interventions, however, can prevent this sequence. To measure the full range of resuscitation success for in-hospital resuscitation, the Utstein task force defines a resuscitation attempt as any effort to restore effective ventilation, oxygenation, and circulation to a patient by the use of any or all of the following methods: defibrillation, cardiac pacing, chest compressions, airway interventions, or intravenous medications. These efforts should be included as subsets in template box 3.
Which patients should be moved farther down the main trunk of the template? Some investigators suggest that the majority of patients who should be moved below box 3 will require combination interventions and are in complete cardiac arrest. The definition of complete in-hospital cardiac arrest is based on the definition recommended for out-of-hospital arrest: absence of a palpable pulse and initiation of chest compressions.24 30 All signs of cardiac arrest, however, such as unresponsiveness and apnea, may not be relevant in all hospital patients. Sedation or anesthesia-induced apnea during artificial ventilation may eliminate the criterion of unresponsiveness. Therefore, the following definition of complete cardiac arrest is recommended:
- Absence of a palpable pulse
- Unresponsiveness due to any cause
- Apnea, agonal respiratory attempts, or artificial ventilation
Template Box 4: False Arrests
BLS or ALS actions not needed
Box 4 provides a place to record false arrests or activations of a resuscitation response for an event that does not represent a true cardiopulmonary emergency. Many events, such as vasovagal episodes, faints, seizures, calls for security help, or a variety of other calls, may precipitate an in-hospital emergency response. Personnel often complete an emergency response form or in-hospital audit form for these events. Several researchers, however, have argued for the need to avoid using near-arrests, false arrests, or response team activations to calculate outcome rates.125 127 128 Although these events document the activity of a hospital's emergency response team, such events should not be used to compare outcomes between hospitals.
Template Box 5: Non-VF/VT
Asystole
PEA
Asystole is defined as complete electrical silence, that is, no cardiac electrical activity can be detected with body-surface electrodes. Electrical activity that is <1 mm in amplitude (at a calibration of 10 mm/mV) should be classified as asystole. The rate division between asystole and PEA (see below) is not well defined. A pulseless rate <10 beats per minute (bpm) was selected as the definition of asystole, but this is an arbitrary figure. So-called "P-wave asystole" was discussed as a possible subset of asystole that should be identified and treated specifically, for example, with cardiac pacing. The task force, however, could reach no consensus on this entity and whether it should be treated differently from asystole without P waves.129 In-hospital patients with asystole may have a prognosis different from that of out-of-hospital patients with asystole130 because in-hospital asystolic patients have a shorter duration of asystole.
PEA is any other non-VF/VT pulseless rhythm apart from asystole. A variety of subcategories of PEA are possible. Narrow complex PEA is thought to have a better prognosis than wide-complex PEA, especially if the PEA rate is rapid (suggesting, eg, hypovolemia). PEA with a slow rate may respond to cardiac pacing. The value of subdividing patients into such categories is unclear, and no specific recommendations were developed.131
Template Box 6: Initial Rhythm VF/VT
VF is defined as disorganized, irregular electrical activity that
produces no appreciable cardiac pumping action. The dividing point
between asystole that should not be shocked and "fine VF" that
should be shocked remains controversial. Utstein I defined VF as
starting when the peak-to-trough deflections on the surface
electrocardiogram are >1 mm in amplitude (at a
calibration of 10 mm/mV) and faster than 150
bpm.24
Studies have shown that a lower percentage of victims of in-hospital cardiac arrest are in VF compared with out-of-hospital patients. Nevertheless, VF is the "rhythm of survival" for most cardiac arrest patients, both in hospital as well as out of hospital. Rapid identification of patients who are in VF must remain a major objective of all resuscitation attempts.
With box 6 the template has arrived at the index point for all subsequent outcome results. Boxes 1 through 4 present important data for setting the context of outcome results. Although a variety of outcome rates can be calculated using the numbers in boxes 1 through 4 as denominators, the task force recommends that box 6 serve as the parent denominator for uniform reporting of adult in-hospital arrests. As discussed below, there are several recommended survival rates for interhospital comparisons. Process gold standards (eg, intervention intervals), however, may prove to be more useful in light of the complicating elements of frequent comorbid conditions and varying levels of severity.
Template Box 7: Never Achieved ROSC
Box 7 displays the true failures of in-hospital resuscitation,
that is, patients who do not respond to resuscitation efforts with even
a brief ROSC.
Template Box 8: Any ROSC
In the out-of-hospital Utstein style, any return of a spontaneous
pulse, detectable by palpation of a central artery (carotid or
femoral), is considered ROSC; no minimal duration for spontaneous
circulation is required.24 This recommendation intends to
capture any possible promising therapy or intervention, even though a
few minimally palpable pulses cannot be considered return of
"spontaneous circulation." Intermittent ROSC refers to patients
who have brief periods of ROSC but who require CPR between these
episodes. Sustained ROSC is defined as the single, continuous presence
of palpable pulses for more than 20 minutes. ROSC lasting longer than
20 minutes may also be used to define when one CPR attempt ends and a
new episode begins.
Arrest may occur in a patient who already has invasive hemodynamic monitoring devices in place. Cardiac output may be detected by techniques such as intravascular pressure monitoring, transesophageal or thoracic echocardiography, or Doppler pulse detection. In these patients, detectable cardiac output should be recorded, even when a pulse is not palpable. Detectable cardiac output without a palpable pulse ("pseudoelectromechanical dissociation") may be associated with improved outcome.132
Template Box 9: Died in Hospital
ROSC 
20 minutes
ROSC >20 minutes but 24 hours
ROSC >24 hours
Box 9 is for recording patients who are successfully
resuscitated but who do not survive to hospital discharge. The task
force recommends three categories of ROSC, each one indicating a longer
duration of sustained, spontaneous circulation: 20 minutes, >20
minutes but 24 hours, and >24 hours. The 20-minutes category is
for patients who are briefly resuscitated; they may experience only a
few seconds of detectable cardiac output. Resuscitation in such
patients cannot be called successful. Patients who live at least 24
hours (ROSC >24 hours) are considered successfully resuscitated (see
below) even if they do not survive to discharge. Patients with ROSC
>20 minutes but 24 hours are an intermediate category. Box 9 was
included in the template to attach greater meaning to sustained ROSC.
Early death in the first day most often occurs as a result of
continuing cardiovascular compromise in patients who
never regain consciousness after CPR.117
Reattempts/rearrests
Template box 9 also presents a means for handling the complicated problem of analyzing patients with multiple arrests during the same hospitalization. If every attempted resuscitation is reported as a separate event, rates of successful resuscitation will be inflated, and rates of survival to hospital discharge will be reduced. This poses the methodological questions of whether the denominator for outcome measurements should be reported per patient or per attempted resuscitation and whether the numerator should be successful resuscitation (ROSC >24 hours) or successful discharge. The following example illustrates this problem.
Two patients experience in-hospital cardiac arrest. The first patient arrests, is successfully resuscitated, and is discharged alive. The second patient has a stormy course, arresting 9 times during a prolonged hospitalization. Personnel cannot resuscitate this patient during the ninth arrest, and the patient dies. The following outcome rates can be calculated:
The task force recommends that all three outcome rates be available. Because some hospitals will have only a small number of patients in these categories, calculation of all possible rates or percentages may fail to adequately express the clinical variability that exists. Confidence intervals improve interpretation of the study data by indicating the degree of uncertainty about an observation. Recent resuscitation research provides a useful illustration of the use of confidence intervals to express outcome data.133
Identify the number of patients in whom
- DNAR status was declared
- Support was withdrawn
- Brain death was established
- Organ donor status was declared
Physicians may treat patients several different ways after a successful resuscitation. Some patients may be given maximum therapy in a concerted effort to help the patient survive to hospital discharge. Alternatively, physicians may, in consultation with family members, reevaluate survivors after the first in-hospital arrest and place these patients in other management categories, which include declaring the patient brain dead or DNAR, withdrawing ventilatory and pharmacological support, and considering whether the resuscitated patient would be a suitable organ donor. Subsequent cardiac arrests in these patients may be "unresisted," as suggested by Longstreth and Dikmen.67 The management of these survivors of an initial cardiac arrest can profoundly change in-hospital survival rates.
Identification of patients for whom support is limited or withdrawn will help answer an important question: Are some differences in survival between centers related to different approaches to the care of patients with severe compromise after resuscitation? From an epidemiological perspective, these patients are eliminated from the survivor numerator. Their failure to survive would adversely reflect on the resuscitation success of the hospital. Because these patients are removed from any possibility of surviving the index cardiac arrest, they should be identified and specifically reported in template box 9.
Template Box 10: Discharged Alive
Total
- Alive at 6 months
- Alive at 1 year
The most frequently used outcome measure for cardiac resuscitation is the rate of survival to hospital discharge. The numerator for this rate is the number of patients discharged from the hospital with spontaneous circulation and ventilation. The task force recommends notation of the number discharged alive, the number alive at 6 months, and the number alive at 1 year.
Unique discharges, such as patients with brain death transferred to another hospital for organ donation, should be reported separately. Such patients do not represent a successful resuscitation. A variety of discharge destinations are possible, ranging from preadmission residence to intensive care custodial facilities. Patients transferred to another hospital for further diagnostic or therapeutic procedures are listed as discharged alive for the hospital where the cardiac arrest occurred. Accounting for these patients, if they die in the recipient hospital, will occur when length-of-survival and functional outcomes are determined (template boxes 11 and 12).
Patients may occasionally be discharged with a spontaneous pulse and respirations yet require intravenous medications such as vasopressors, antiarrhythmics, and cardiac inotropes, or intermittent advanced ventilatory support. Whether such patients merit reporting as successful survivors of in-hospital cardiac arrest is debatable. Common sense is needed. Exceptional patients may fall so far outside the usual definitions that they will need to be reported separately.
Template Box 11: Died Within 1 Year of Discharge
Template Box 12: Alive at 1 Year
Once a patient is discharged alive from the hospital, outcome
auditing requires date of death (if death occurs during the first year
after discharge) or confirmation that the patient is alive after 1
year. These data do not flow routinely into the hospital where the
index arrest occurred. Researchers must therefore establish an
additional data collection system, such as telephone calls and
surveillance letters; postcards and questionnaires to care facilities,
patients, relatives, and physicians; death certificate review; and
periodic home visits. The key, according to researchers who have
published the best data on length of survival, is a single, dedicated
research associate who establishes a close relationship with the
survivors and their families and contacts them at regular
intervals.105 134 In the United Kingdom, the Office of
Population and Census prospectively tracks people who meet specified
criteria (eg, survivor of cardiac arrest) and notifies researchers of
the date and certified cause of death when death occurs.
Problems arise when patients are transferred (from hospital to nursing facility to home and back again); when patients or their families change hospitals and physicians; when patients relocate (to other addresses, cities, even states and countries); when telephone calls or surveillance letters are not returned; and when there is a lack of cooperation among the physician, patient, and family.130 135
The date of death for arrest survivors who die after discharge should be available through death certificates. Gaining access to and tracking death certificate data may be a problem in some regions, states, and countries. In the United States, social security numbers are the indexing codes used in most death-certification programs. This number, however, is often not recorded in hospital records or is not available to researchers because of patient confidentiality concerns.
The survival data can be presented in several valid ways. A simple percentage (number alive at 1 year over number discharged alive from the hospital) is the most straightforward calculation. However, survivors to hospital discharge are not released at one point in time; a hospital's successful discharges are staggered over many weeks and months. Therefore, by definition, all survival-to-1-year data will be 1 year out of date because allowance must be made for the last individual discharged alive to have a chance to live 1 year.
Most survival studies solve this problem by using some method of calculating probability of survival to 1 year. These probabilities then must be adjusted for attrition, death, relocation and transfer, inability to follow up, loss to follow-up after enrollment, lack of information, or lack of cooperation. Newly discharged-alive patients enter the study survivors group upon discharge. They remain in that group until "removed" from the cohort by one of the events noted above (death, loss to follow-up, etc). Kaplan-Meier survival curves are the most widely used statistical means for presenting these data. (For more detailed information, refer to Kaplan and Meier.136 )
Another recommended statistical approach to expressing survival is a modification of the life-table method adopted by the United Kingdom Resuscitation Council in BRESUS.22 In that study survival was viewed from two perspectives. First, beginning at the time of the arrest, investigators determined how many people would live through subsequent time periods. Survival results can be reported from either single or overlapping periods. For example, the BRESUS investigators observed the following survival rates: 71.7% from the arrest to 24 hours, 62.7% from 24 hours to discharge, 71.8% from discharge to 1 year, 32.4% from 24 hours to 1 year, and 12.5% from arrest to 1 year.22 In the second approach, the BRESUS investigators asked how many attempted resuscitations are required to produce 1 survivor at 1 year. The results demonstrated that for every 1-year survivor there were 8 attempted resuscitations, 3 initial survivors, 2 who survived the first 24 hours, and 1.5 who left the hospital alive.22
The Utstein Task Force recognizes the need to establish a body of published survival rates of in-hospital cardiac arrest that use comparable definitions of survival. Because the BRESUS report is the largest existing report of in-hospital survival rates, it is advantageous for other projects to use the BRESUS definitions. The Utstein Task Force therefore recommends that researchers use the data from template boxes 9 through 12, combined with the life-table method to allow for losses to follow-up,22 to calculate the following survival rates:
- Percent survival from the arrest to at least 24 hours (ROSC >24 hours)
- Percent survival from successful resuscitation (ROSC >24 hours) to 1 year
- Percent survival from arrest to discharge
- Percent survival from discharge to 1 year
- Percent survival from arrest to 1 year
For each survivor alive at 1 year, calculate the following:
- Number of attempted resuscitations
- Number of initial survivors
- Number surviving first 24 hours
- Number leaving the hospital alive
Template Box 13: Functional Outcomes
The functional outcome (neurological and psychological status)
should be measured at the following times:
- Hospital discharge
- 6 months to 1 year
The dashed line to box 13 indicates that assessment of outcome is completed only by providing a measure of functional outcome. Ideally, researchers should record the best neurological and psychological status achieved after resuscitation. Research, however, has not established the best measures of quality-of-life or functional outcomes after cardiac arrest.105 106 116 130 131 The optimal time intervals for performing functional assessments are unknown.118 The task force recommends, as a minimum, assessment of functional outcome near the time of hospital discharge and again at 6 months to 1 year. Practical concerns support general use of the CPC score, whereas for more focused research projects additional measures are needed.116 Some research suggests that patients reach their maximum level of recovery from the original anoxic or ischemic event of the cardiac arrest between 6 and 12 months.118 After that point, patients begin to experience mortality and morbidity because of age and preexisting illnesses.106
|
Data to Collect on Individual Patients |
|---|
Table 1
lists the data on individual patients that the task force
recommends be collected. These data will support analyses for
intrahospital audits and interhospital comparisons and will allow
completion of the in-hospital Utstein-style template (Fig 2). This
approach was pioneered in the first Utstein-style guidelines for
reporting out-of-hospital cardiac arrest.24 30 However,
the task force acknowledges a conceptual debt to the uniform data sets
designed by Spaite et al137 for the National Highway
Traffic Safety Council.
|
Unfortunately, no single cardiac arrest audit form can capture all the
recommended data. The primary problem emanates from the methodological
challenge of data synchronization. This problem is illustrated in Fig 3, which demonstrates that researchers can gather
complete resuscitation data about an individual only by moving back and
forth in time.
|
An individual patient first becomes of interest for data collection when he or she suffers an in-hospital cardiopulmonary emergency. At the time of the resuscitation attempt, personnel record only immediate hospital and patient data about the resuscitation (eg, time of defibrillation). Immediately after resuscitation, patient variables can be gathered by reviewing the patient's hospital records. Outcome data (eg, date of death) can be collected by a prospective surveillance system, but this requires linking the clinical information with the surveillance system. Prospective surveillance may incorporate use of telephone calls, physician office records, and death certificates. Most functional outcome measures will require patient examinations or face-to-face interviews. The problem of multiple records stored and analyzed over time has been reduced significantly by the widespread availability of personal computers and easy-to-use data management software. Nevertheless, considerable advance thought and planning is needed to conduct comprehensive analyses of in-hospital resuscitations.
|
Gold Standard for Outcome Comparisons |
|---|