A Case for the Use of Validated Physiological Mortality Metrics to

Transcription

A Case for the Use of Validated Physiological Mortality Metrics to
AACN Advanced Critical Care
Volume 26, Number 1, pp. 13-22
© 2015 AACN
A Case for the Use of Validated
Physiological Mortality Metrics to
Guide Early Family Intervention in
Intensive Care Unit Patients
Molly F. Searl, RN, BSN, MSN, CCRN
ABSTRACT
In the current health care climate a large portion of health care dollars are spent in the
final months of life, so ensuring that care
provided is in line with the wishes of patients
and their families is more critical than ever.
On the one hand, surviving families often
report that they wish they had been given
prognostic information earlier and that, in
retrospect, they would have made different
treatment decisions if they had been given
prognostic information. On the other hand,
providers often are reluctant to discuss prognosis for various reasons, not the least of
which is the inherently uncertain nature of
prognostication. To address this issue, this
article reviews pertinent literature about
provider reticence, family preference, and
the move toward palliative care and includes
a discussion of the various validated
mortality-prediction models available. A case
is made to use those validated metrics to
guide early discussions of palliative or endof-life care for patients who are critically ill. A
suggested checklist to facilitate inclusion of
prognosis discussions in family meetings is
included as well as a case study to illustrate
the problem, current practice, and a model
for improvement.
Keywords: critical care communication, end
of life, mortality prediction scores, palliative
care, prognosis
vasopressor support before he could be taken for
confirmatory computed tomography scan. A
bedside echocardiogram showed profound rightsided heart failure, with a nearly akinetic right
ventricle. When his son arrived on postoperative
day 1, his first question to the health care team
Mr S is a 96-year-old man with a recent history
of recurrent falls and consequent orthopedic
fractures. Before his most recent hospital
admission, he fell and sustained a left-sided
femur fracture. While being evaluated for open
fixation of this fracture, he had chest pain and
was diagnosed with a non–ST-elevation
myocardial infarction for which he underwent a
cardiac catheterization on hospital day 2. After
being cleared for the orthopedic operation by
cardiologists, he went to the operating room
on hospital day 5 for a left femur fixation.
Initial postoperative clinical findings were
concerning for a pulmonary embolism, and on
his first night after the operation, his clinical
condition deteriorated, requiring intubation and
Molly F. Searl is Critical Care Transport Nurse, The Johns Hopkins Hospital, 1800 Orleans St, Baltimore, MD 21287
([email protected]).
The author has no financial disclosures or conflicts of interest
to report. The views expressed herein are those of the author
and do not necessarily reflect the views of The Johns Hopkins
Hospital.
DOI: 10.1097/NCI.0000000000000063
13
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 13
09/01/15 8:06 AM
SEARL
W W W.A ACNA DVA NCE D CRIT ICA LCA RE .COM
can facilitate a change from strictly curative
to either a blended curative-palliative or a
comfort-only model of care, both of which
have been shown to possibly extend life and
improve quality of life for both patients and
caregivers.1,7,13
Unfortunately, fewer options currently are
available to reduce the stress providers feel
when providing a poor prognosis.6 Education
and debriefings have been proposed as a means
to reduce these feelings of stress, but few data
are available on the effectiveness of these interventions.7 With regard to feelings of failure,
they are relatively typical for providers who do
not have training in either palliative or end-oflife (EOL) care,8 although more attention has
been paid in recent years to the idea that the
quality of a patient’s death should be given as
much credence as the quality of his or her
life.1,14 One solution proposed and widely
accepted to manage both stress and feelings of
failure is the integration of palliative care,
through the inclusion of specially trained providers, into ICU teams.15 Alternatively, critical
care providers can be given basic training on
how to discuss and integrate palliative care into
their daily practice. The feasibility of such
interventions depends on the size of the institution, funding availability, and staff buy-in15 and
is beyond the scope of the current discussion.
Finally, providers cite unreliable predictors
of prognosis as a reason for not initiating discussions.7 Although no provider can predict
with absolute certainty when a patient will die,
metrics are available that can help provide a
scientifically based prediction of in-hospital
mortality for patients who are critically ill.16
Few data are available on the current accuracy
or frequency of prognostic discussions in ICUs.
Breslow and Badawi17 found that, despite recommendations, only 10% to 15% of ICUs use
mortality-predicting metrics. The recommendations cited suggest the use of prognostic metrics as indicators of quality improvement and
care standardization, not as guidelines for the
treatment of individual patients.17 Clinically,
the general trend tends to be that prognostic
information, when it is presented, is based not
on metrics but on clinician experience or opinion, which can be quite varied.5,17 Research is
not consistent with regard to whether discussion of prognosis decreases ICU resource use18
or length of stay,18 or increases the number of
do-not-resuscitate orders,4 but when prognosis
is discussed, variation in the information
was, “Can my dad make it through this?” The
answer was, “We just don’t know.”
C
ommunication between physicians and
patients or their families in the critical
care setting has been well documented as an
area in need of improvement.1–5 Also well documented is the difficult nature of discussing
prognosis, specifically the disconnect between
a provider’s willingness to discuss prognosis6–8
and the desire of patients and families to be
given accurate information about prognosis.9–12 The purpose of this article is to make a
case for the use of validated physiological
prognostic metrics to aid in the discussion of
prognosis in patients who are critically ill, and
thus improve communication between the
health care team and the families of patients in
the intensive care unit (ICU). By using metrics
as a guide, providers could more easily initiate
early implementation of palliative or comfort
care for patients who are critically ill, many of
whom often undergo aggressive treatment
until the final days of life.1,13 As a means of
illustration, the case study described earlier is
both an example of current practice and a
model for improved communication.
The Difficulty in Discussing
Prognosis
Relatively few studies have investigated the
reasons why critical care providers choose not
to provide prognostic information to patients
or families. Those that have been conducted
tend to report 3 recurring themes: (1) the belief
that providing a poor prognosis will cause
patients to lose hope and potentially deteriorate more quickly than if the information had
been withheld6; (2) the concern that discussions of prognosis, especially when the estimated survival time is short, cause a large
amount of stress among providers,6 often making them feel as if they have failed in their duty
to care for their patient8; and (3) the idea that
prognosis is uncertain and there is no reliable
way to predict the course of any individual’s
disease with absolute certainty.7,8
Regarding clinical deterioration, studies
have reported that providing prognostic
information to patients, even if the prognosis
is grim, actually fosters hope, because the
patients have a realistic notion of their disease
trajectory and can trust their providers to be
honest with them, regardless of the situation.7
Early and frank discussions of prognosis also
14
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 14
09/01/15 8:06 AM
VOL UME 2 6 • N U MBER 1 • JANUARY- M ARCH 2015
VA LIDAT E D P H YSIOLOGICA L M ORTA LIT Y M E T RICS IN T H E ICU
presented to patients and families has been
cited as a significant barrier to effective EOL
care in the ICU2 and can lead to ineffective use
of palliative care,19 increased caregiver stress,9
and poor satisfaction as reported by surviving
family members.3
The difficulty in providing prognostic information in the critical care setting is multifaceted but can be generally summarized by
recognizing that clinicians, be they doctors,
nurses, nurse practitioners, or any other members of the health care team, have difficulty
reconciling the seemingly mutually exclusive
mandates to save lives and provide effective
palliative care.19 Often, the treatments with the
greatest likelihood of success are those that
cause distressing symptoms in patients,20 and
those symptoms often are the aspects of EOL
that cause the most dissatisfaction among surviving families.3
pare can be addressed through early initiation
of family meetings,13 a key part of which
should be—based on family surveys—a discussion of prognosis.9 Qualitative studies have
shown that the desires described earlier, specifically the desire to have discussions of prognosis, contribute significantly to improving
family satisfaction, which is increasingly recognized as a measure of successful treatment.21
Families have reported that their treatment
decisions, made on behalf of their loved ones,
may have been different had they been
informed of the objective prognosis early in the
disease course,10 but data are mixed. Both the
seminal Study to Understand Prognoses and
Preferences for Outcomes and Risks of Treatments (SUPPORT)4 trial and newer research by
Daly and colleagues18 report no change in EOL
decision making after the initiation of structured family communication that included discussions of prognosis. No information is
available about the source—provider experience or physiological metric—of that prognostic information, although Daly and colleagues18
report that Acute Physiology and Chronic
Health Evaluation (APACHE) III data were
collected and scores were recorded as part of
the patient demographic information.
Note that mortality scores report predicted
survival, but the perception of what survival
actually means can be very different between
families and providers.10 Specifically, families
often consider survival to mean a return to previous functional status,12 whereas survival to
providers often means “alive at discharge,”
with less focus on the postillness functionality
of the patient.12 In fact, with the advent and
proliferation of life-sustaining treatments,
patients and families tend not to understand
that medical intervention can—on numerous
occasions—only serve to prolong the dying
process12 and is often the source of many distressing symptoms20 identified by family members as a significant source of regret after the
death of their loved one.3 Providers must recognize the need to reduce suffering and not
allow the potential adverse effects of treatment
of distressing symptoms to prevent such treatment.20 As an example, a provider should not
expect a patient to tolerate pain (a distressing
symptom) to prevent or manage hypotension
(a physiological condition), because treatment
of the symptom (pain) may exacerbate the condition (hypotension), without first discussing
treatment options and possible consequences
“Mr S cannot possibly survive this” was the
general consensus on the part of not only the
ICU team caring for him but also the primary
surgical team on postoperative day 1. This
view was expressed by multiple care providers during evening rounds. Mr S’s son was
present at the bedside throughout the day,
although he was not present during rounds
when this discussion occurred. When providers entered the patient’s room after rounds,
Mr S’s son asked, “How’s Dad doing?” to
which the response was, “He is very sick. He
needs a lot of help maintaining his blood pressure and breathing right now.” No discussion
of data-driven prognostic information was
offered.
The Need to Discuss
Prognosis
Family Perspective
When surveyed, families of critically ill patients
report that their main concerns—after curing
the underlying disease—are to ensure that their
loved one does not suffer unduly,6 that distressing symptoms are addressed and treated in
a timely fashion,20 and that they be allowed to
share in decision making9 in an effort to prepare themselves for the possibility that their
family member might not survive.10 Issues of
suffering and symptom relief can be addressed
with the inclusion of palliative care in all critical care settings, regardless of prognosis,15 but
this, again, is beyond the scope of this discussion. Shared decision making and time to pre15
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 15
09/01/15 8:06 AM
SEARL
W W W.A ACNA DVA NCE D CRIT ICA LCA RE .COM
with the patients and families.20 Avoidance of
suffering, or improved quality of life, has been
cited as a primary goal for families and patients
during illness,22 and research supports the
notion that morbidity rather than mortality is
a more accurate term to describe the concerns
of patients and families.23
Although many researchers have compiled
data on objective mortality-prediction measures, data reflecting morbidity are, by necessity, based on surveys of survivors and their
families. These surveys, collectively known as
health-related quality of life surveys, collect
subjective data on a number of variables,
including functional status, anxiety, and
depression, as well as objective measures, such
as delirium and pulmonary, renal, and cognitive function.23 When combined, these measures give a picture of functional status, which
is often decreased, after critical illness.23 Data
report that patients who require longer and
more aggressive care are likely to experience
greater functional dependence after critical illness.24,25 Patients with comorbid conditions on
admission and patients who are older or
require ventilator support for longer periods of
time are at highest risk for postillness sequelae.24 These same physiological variables also
are used to calculate mortality scores; thus,
patients with higher mortality scores on admission are more likely to require aggressive care26
and, therefore, are more likely to experience
long-term functional consequences. Therefore,
logically, predictors of mortality—that is,
prognostic metrics—can be extrapolated to
predict postillness morbidity for those patients
who survive their critical illness. This postillness morbidity is, as mentioned earlier, generally the most important factor for patients and
their families and often is not discussed during
family meetings.27
the provider is not completely sure.9 In fact, the
tendency of providers to tell families that it is
best to “wait and see” before discussing prognosis,11 or the fact that different providers
often present different prognoses,2 is actually
detrimental to families and limits their ability
to process information and make appropriate
treatment decisions.10 Conflicting opinions
about prognosis and efficacy of treatment
often lead to staff burnout related to caring for
patients who are chronically critically ill,2 and
the hesitancy on the part of providers to discuss prognosis has been shown, in some cases,
to actually contribute to a decrease in frequency of family conferences as length of stay
increases, when illnesses become more complicated, and the need for realistic discussions of
prognosis becomes more pressing.5
Postoperative day 2: Mr S has not improved.
His liver function has decreased, demonstrating catastrophic liver injury; his kidney function has deteriorated; and he still requires
maximal vasopressor and ventilator support.
His son remains at the bedside, and frequently
asks how his father is doing. During a discussion with his father’s physician, he relayed the
story of his mother’s death approximately
1 year before. She had suffered a relatively
minor injury and had developed sepsis as a
result. He stated, in passing, that he did not
want his father to suffer like that, relying on
machines to keep him alive. Again no discussion of prognosis was offered.
Arguably, a disconnect exists in perception
that leads to a demonstrable lack of communication between providers and families.3 That
disconnect can be traced, at least partially, to
discomfort on the part of providers to discuss
prognosis in concrete terms, because they are
uncomfortable in providing such information
without supporting evidence, because there are
conflicting opinions on prognosis, or because
they are justifiably uncomfortable with or
undertrained in how to provide potentially devastating news to patients and their families. The
question then becomes whose opinion on prognosis should be used to guide discussions.
Should it be the primary service attending, in
the case of a surgical ICU patient, who has seen
thousands of postoperative liver cancer patients
as an example? Or should it be the intensivist
who has cared for thousands of critically ill
individuals regardless of the underlying pathology? Or should it be the palliative care team,
Provider Perspective
Perhaps the most substantial argument against
discussing prognosis, from a provider’s perspective, is uncertainty; even the most accurate
prognostic metrics are variable and designed
specifically to aggregate mortality information
for groups rather than predict mortality for
specific patients.17 Research supports the idea
that providers are generally less likely to provide specific information about prognosis than
any other aspect of health care,6 even though
family members report that they generally
would prefer a discussion of prognosis, even if
16
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 16
09/01/15 8:06 AM
VOL UME 2 6 • N U MBER 1 • JANUARY- M ARCH 2015
VA LIDAT E D P H YSIOLOGICA L M ORTA LIT Y M E T RICS IN T H E ICU
whose job is to focus care on comfort first,
either as an alternative to or in collaboration
with curative efforts? In an attempt to remove
opinion from the already-complicated discussion, the use of valid and proven prognostic
metrics is a reasonable and relatively easy
solution.
Table 1: Advantages, Limitations, and
General Considerations of 3 Prognostic
Metricsa
Advantages
All have well-established discrimination and
calibration in general ICU patients
APACHE, SAPS, and MPM have been validated
in large numbers of clinical trials spanning
all critical care settings (ie, surgical, medical,
neurological)
Prognostic Metrics
Various prognostic metrics that include both
general prognostic scores and organ dysfunction scores are available for providers.28 For the
purpose of this discussion, only general prognostic metrics are reviewed; the most commonly
used metrics are the APACHE, Simplified Acute
Physiology Score (SAPS), and Mortality Prediction Model (MPM).29 These 3 metrics have been
found to have good discrimination—the ability
to identify ICU patients with the highest risk of
death—and good calibration—the extent to
which the predicted outcome matches observed
outcome.17 In brief, these prognostic metrics
were developed in an effort to better categorize
disease severity and to standardize research in
the critical care setting, and as a basis for comparing care across ICUs.16 See Table 1 for a general appraisal of advantages and limitations
common to all 3 metrics. A more detailed discussion of each metric follows.
Based on physiological data that, when combined, provide accurate mortality predictions
Limitations
Not well studied for specific ICU patient populations (eg, AIDS, postpartum) or in ICUs with
specialized treatment modalities (eg, ECMO)
Some disease processes have expected trajectories that can change scores based on when
they are calculated (eg, expected low GCS with
new SAH patients could improve with time)
Lead-time bias—the change in score based on
treatment before data collection—can alter
scores, especially in patients who have been
transferred from one facility to another
General considerations
All scores can be affected by improvements in
medical technology, implementation of best
practice recommendations (eg, early antibiotics
for suspected or confirmed sepsis), or changes
in practice (eg, new providers rotating through
the ICU)
Acute Physiology and Chronic
Health Evaluation
Originally proposed in 1981, the APACHE system was designed to provide an objective classification of risk in critically ill patients.30 The
author of this study acknowledged the inherent
reluctance on the part of physicians to discuss
prognosis for individual patients, so this particular tool was designed to refer to groups of
patients with similar physiological variables,
not to guide individual treatment decisions.30
As a result of polling experts for the variables
deemed “important,” Knaus and colleagues30
derived a weighted classification system that
allowed comparison of an individual patient to
a group of similar patients to predict mortality.
This original scoring system, as with all such
prediction metrics, was found to have less discrimination over time, largely due to advances
in medical technology. Consequently, the
APACHE score has undergone many revisions,
the most current being the APACHE IV, which
includes measures of length of stay and incorporates logistical regression to provide users with
a predicted mortality score as a percentage.31
Abbreviations: AIDS, acquired immunodeficiency syndrome;
APACHE, Acute Physiology and Chronic Health Evaluation; ECMO,
extracorporeal membrane oxygenation; GCS, Glasgow Coma
Scale; ICU, intensive care unit; MPM, Mortality Prediction Model;
SAH, subarachnoid hemorrhage; SAPS, Simplified Acute
Physiology Score.
a
Based on data from Kelley.16
The physiological values required to calculate
an APACHE IV score for Mr S are listed in
Table 2; values included therein reflect the
worst values in the first 24 hours as stipulated
by the scoring system. A primary diagnosis
that led to ICU admission also is required for
an APACHE score.17 Previous iterations of the
APACHE were limited by their proprietary
nature, meaning that hospitals were required
to purchase costly software programs to access
the information; however, APACHE IV is now
available online and free of charge at http://
www.apachefoundations.cernerworks.com.31
Other freely available online severity scoring calculators (http://clincalc.com) report APACHE II,
17
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 17
09/01/15 8:06 AM
SEARL
W W W.A ACNA DVA NCE D CRIT ICA LCA RE .COM
Table 2: Raw Data Used to Calculate Mr S’s Mortality-Prediction Scoresa
Patient Data
Temperature, °C
Heart rate, beats/min
Respiratory rate, breaths/min
On Admission
24 Hours After
Admission
36.3
35
93
107
24
30
72/37 (49)
120/49 (66)b
150
0
White blood cell count, 10 /μL
29.7
31
Hematocrit, %
27.2
28.2
Serum sodium, mEq/L
143
142
Serum potassium, mEq/L
3.8
4
Serum bicarbonate, mEq/L
18
19
BUN, mg/dL
26
34
Serum creatinine, mg/dL
0.91
1.67
Serum albumin, g/dL
2.1
1.6
BP (MAP), mm Hg
Urine output, mL/d
3
Serum bilirubin, mg/dL
1.1
2.5
Serum glucose, mg/dL
110
139c
GCS
3T
3T
ABG FIO2, %
100
80
ABG PaO2
166.5
93.7
ABG PaCO2
21.5
29.3
7.2
7.4d
APACHE IV (predicted mortality), %
82.5
87.9
SAPS II (predicted mortality), %
88.9
89.7
MPM II (predicted mortality), %
90
87.3
ABG pH
Sources for predicted mortality data
APACHE IV: http://www.apachefoundations.cernerworks.com
SAPS II: http://clincalc.com/IcuMortality/SAPSII.aspx
MPM admission: http://www.sfar.org/scores2/
mpm2_admission2.php
MPM 24 hours: http://www.sfar.org/scores2/
mpm2_24_48_722.php
Abbreviations: ABG, arterial blood gas; APACHE, Acute Physiology and Chronic Health Evaluation; BP, blood pressure; BUN, blood urea
nitrogen; FIO2, fraction of inspired oxygen; GCS, Glasgow Coma Scale; MAP, mean arterial pressure; MPM, Mortality Prediction Model; PaCO2,
partial pressure of carbon dioxide; PaO2, partial pressure of oxygen; SAPS, Simplified Acute Physiology Score.
a
None of the scores has a caveat for the use of medical support beyond ventilators (accounted for in the ABG values) or sedation medication
(accounted for in the GCS). The use of vasopressors to sustain blood pressure or insulin to control blood glucose is not noted and thus does
not change the weight of any individual scores. The supportive measures noted for some values in the 24 Hours After Admission column are
for the edification of the readers and are meant to clarify that some of the values presented, while they appear improved after 24 hours, are
in fact with more support than previously.
b
Norepinephrine (Levophed) infusion at 0.12 mcg/kg per minute and vasopressin infusion at 0.04 units per hour.
c
Insulin infusion at 5 units per hour.
d
Sodium bicarbonate drip at 150 mL/h.
18
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 18
09/01/15 8:06 AM
VOL UME 2 6 • N U MBER 1 • JANUARY- M ARCH 2015
VA LIDAT E D P H YSIOLOGICA L M ORTA LIT Y M E T RICS IN T H E ICU
SAPS, and other organ-specific mortality scores.
APACHE scores also require a large amount of
clinical data, and without automatic chart
review processes, data entry can be labor intensive. In addition, when data are not available,
the software assumes a normal value, which can
potentially lower the calculated score, which
was a conscious choice made by the original
APACHE authors based on the assumption that
if a value is not measured the providers at the
time did not suspect it would be abnormal.30
Being the most recent iteration of the APACHE
scoring system, the APACHE IV score calculated via the above website is included for Mr S.
and at least one study found it valid in a rural
community hospital.35 A major difference
between MPM and SAPS/APACHE is that the
variables in MPM are binary, that is, marked as
present or not present, thus significantly reducing the amount of time and specificity of data to
be collected.16 Another difference is that the
MPM produces a predicted mortality percentage, rather than a score that is then translated
into a percentage.
Based on clinical data for Mr S, his initial
APACHE IV score was 146, with a predicted
mortality of 82.5% for this hospitalization.
Data are included in Table 2.
These metrics have limitations on their
effectiveness in certain patient populations or
under certain conditions,16 so clinical judgment
should guide the timing of score-driven family
meetings in these circumstances. See Table 1
for a summary of these issues. Regardless of
the metric chosen, the basic underlying supposition is that a set of physiological variables,
when combined, can point toward a potential
for mortality. Although none of the authors
who designed or validated these scoring metrics advocates their use in guiding treatment
for specific patients, their use in reducing interprovider disagreement about prognosis cannot
be denied.
Based on clinical data for Mr S, his initial MPM
predicted mortality rate was 90% for this hospitalization. Data are included in Table 2.
Simplified Acute Physiology Score
From the original APACHE data, Le Gall and
colleagues32 created the SAPS, which was based
on the same principles as APACHE but sought
to streamline and simplify the process by
decreasing the number of variables required.
As with APACHE scores, SAPS has been validated throughout critical care populations and
has undergone 3 revisions, the most recent of
which, SAPS 3, includes customized variables
for geographic regions and has helped increase
discrimination.28 Perhaps the most important
difference between SAPS and APACHE is the
recognition on the part of Le Gall and colleagues33 that selecting a single primary diagnosis for patients in the ICU with complicated
health issues and often with comorbidities is
difficult, requiring provider judgment and
opinion to determine the primary diagnosis.
By postoperative day 7, Mr S continued to
require full ventilator support and multiple
vasopressors, and his kidney function had
deteriorated to the point where he needed
dialysis—a treatment deemed futile by the
Renal Service. While his cardiac function and
laboratory tests showed some improvement,
his respiratory status was complicated by a
multidrug-resistant Proteus pneumonia, and
he never regained consciousness despite a lack
of sedation medications. Attempts were made
by the ICU team to procure a do-not-resuscitate order during the day on postoperative day
6, but again with no concrete discussions of
prognosis, Mr S’s son continued to cling to the
hope that his father might recover. Frustrated
with the discussions he had with the ICU team
and his perception that the team had “given
up without reason,” Mr S’s son requested that
the patient be transferred to another facility,
which occurred on postoperative day 8.
Based on clinical data for Mr S, his initial
SAPS score was 75, with a predicted mortality of 88.9% for this hospitalization. Data are
included in Table 2.
Mortality Prediction Model
Finally, in an attempt to remove expert opinion
and therefore clinician bias, Lemeshow et al34
created the MPM, which derived variables and
weights through statistical regression. The
MPM is designed to have data entered at 2 predetermined times, on admission and at 24 hours
after admission. As with the other 2 scores,
MPM has been validated and found to be accurate across a large number of ICU populations,
Recommendations
Intensive care practitioners have a duty to communicate honestly and openly with patients and
19
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 19
09/01/15 8:06 AM
SEARL
W W W.A ACNA DVA NCE D CRIT ICA LCA RE .COM
their families about disease prognosis. Rather
than continue with the current practice of basing such discussions on individual provider
preferences or experiences, validated physiologically based prognostic metrics should be considered. Best practice encompasses accurate
diagnosis, effective and timely use of available
treatments,26 and discussions of palliative care;
although palliative care measures will not affect
mortality scores, they will affect satisfaction
scores from survivors, which are given nearly
equal weight as benchmarks for successful
care.21
Although it is ideal, frequent contact between
patients and their families and providers is often
cost and/or time prohibitive.7 For providers to
effect the most positive change, all patients should
receive a mortality prediction score on admission
to the ICU. The scores presented here have been
shown to provide similarly accurate information,
and the use of one score over another should be
left to the discretion of the individual institution
on the basis of staff availability for data entry and
common clinical practice. Given the research on
effective implementation of family meetings and
palliative care, the best practice option would be
to integrate a multidisciplinary palliative care
team in any such meeting.15 As a secondary, and
potentially more feasible, option, specific providers could be trained to have discussions about palliative care and prognosis. If neither of these
options is available, any provider can access any
of the aforementioned prognostic metrics and
open a dialogue with families based on the mortality score.
As for which patients need family conferences, those patients with higher severity
scores (60% or greater), and therefore a higher
probability of mortality, should have a family
conference scheduled within the first 48 hours
if possible, although timing could change for
individual patients with certain illnesses or
injuries; see Table 1. For patients with scores
in the midrange (30%-60%), family conferences should be initiated within 72 hours of
admission. For patients with low mortality
scores (< 30%), who have the highest likelihood of returning to function, family conferences could be delayed or even forgone
completely. Lead-time bias, or the effect of
treatments initiated before data collection, can
affect mortality-prediction36 scores, but this
should not—as a rule—prevent score calculation or early family meetings. In an effort to
address concerns raised by bedside nurses,
specifically a lack of protocols for discussions
of prognosis, and the view that families are
often given conflicting information about
prognosis,2 a checklist incorporating the suggested prognostic levels above is included in
Table 3.
As with most physiological variables, trends
are more significant than individual numbers, so
periodic recalculation of prognostic scores is
suggested. If the MPM model is used, an automatic recalculation is available, in that online
calculators have options for admission data and
data at 24, 48, and 72 hours, respectively.
Although APACHE and SAPS use the worst values collected in the last 24 hours, no provision is
made for recalculation. The checklist in Table 3
includes suggested decision points that prompt
Table 3: Suggested Checklist for
Implementing Family Meetings Based
on Mortality Predictions
Patient Name:
Date:
Time:
Admission APACHE II Score/
Predicted Mortality:
Family meeting indicated for
> 30%:
Within 48 hours (predicted
mortality > 60%)
□
Within 72 hours (predicted
mortality 30%-60%)
□
Meeting scheduled for/with:
Advance directive/living will?
Reviewed: □
□ YES
□ NO
□ YES
□ NO
Reevaluate in 72 hours if patient is still in the intensive
care unit or with worsening
clinical conditiona
Repeat APACHE II Score/
Predicted Mortality:
Date/Time:
Family Meeting
Indicated?
(predicted mortality
> 30% or change in
score ≥ 15%)
Meeting Scheduled for/with:
Abbreviation: APACHE, Acute Physiology and Chronic Health
Evaluation.
a
Changes in condition include but are not limited to: addition of
vasopressors, emergent surgery, increase in oxygen requirements,
unplanned intubation, decrease in mental status, and cardiac/
respiratory arrest.
20
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 20
09/01/15 8:06 AM
VOL UME 2 6 • N U MBER 1 • JANUARY- M ARCH 2015
VA LIDAT E D P H YSIOLOGICA L M ORTA LIT Y M E T RICS IN T H E ICU
recalculation of scores at a specified time or with
major changes in patient status, as defined on
the checklist. Table 4 includes a suggested script
for providers to discuss prognosis based on metrics, using Mr S as an example.
Summary
In 1998, the Institute of Medicine defined a good
death as “one that is free from avoidable distress
and suffering for patients, families, and caregivers; in general accord with patients’ and families’
wishes; and reasonably consistent with clinical,
cultural, and ethical standards.”14(p1) Based on
that recommendation, and the recognition that
EOL in hospitals, especially in critical care, can
be unduly and unnecessarily distressing to families,10 a current trend in health care is to provide
quality and timely palliative care.1 Considering
these factors, practitioners must initiate discussions of palliative care early in a patient’s critical
illness, regardless of prognosis,20 but especially
when the probability of death is high.10
Arguments can be made that prognostic metrics are not fail-safe or meant to guide treatment
decisions for individual patients, but in light of
the fact that families often desire prognosis
information,2 even if that information is uncertain,9 and given the frequent disagreements
between providers on prognosis,2,19,22 a case can
be made that using these metrics to guide early
discussions of palliative or EOL care could (a)
improve family satisfaction with the dying process in hospitals,20 (b) give families more time to
comprehend and process the nature of their
loved one’s prognosis,10 (c) decrease staff stress
when caring for a patient that “everyone knows
is going to die,”2 and most importantly, (d)
improve the quality of health care at EOL.
The suggestions presented here should be
subjected to academic scrutiny, but the purpose of this article is to raise awareness of a
practice deficit and offer a solution. Likewise,
the purpose is not to advocate for the withdrawal or restriction of life-sustaining therapies for patients. Nor is it to advocate the use
of prognostic metrics to arbitrarily guide care
for individuals. Instead, given the difficult
nature of both prognostication and discussions of EOL, this article simply suggests the
use of validated physiological metrics to aid
in discussions with families and patients in
the critical care setting about goals of care. By
using established objective prognostic tools,
clinicians can ensure that families are given
the same information about prognosis, and
by initiating discussions about the family’s or
the patient’s wishes early, providers can help
families prepare for poor outcomes. Most
important, with improved communication
comes improved quality of care and, for some
patients, improved quality of their death.
Table 4: Alternate Script Illustrating Use
of Mortality-Prediction Scores in Family
Meeting
Provider:
Mr S, your father’s condition is very
serious. He couldn’t breathe on his
own so we had to put in a breathing tube. His blood pressure is very
low, so we started him on medications to keep it up. And this morning, we did an echocardiogram,
a test that shows us how well his
heart is beating, and it showed that
half of his heart isn’t working.
Family:
Can my dad make it through this?
Provider:
In a man of his age, with all of the
problems he has now, there is an
83% chance of death. We have tools
that help us predict if a person will
survive a hospital stay. Now, these
tools are not 100% accurate. But
they do tell us that 83% of people
with your father’s issues will die.
Family:
So it’s not for sure he’ll die? He was
healthy and independent until this
happened.
Provider:
No, we can’t say 100% that he will die,
but it is very likely that he will not
survive this, and that if he survives,
he likely won’t be able to get back
to his independent life. Did you
father have an advance directive or
a living will?
Family:
No, nothing like that.
Provider:
Do you know if he’d want to be kept
alive on machines?
Family:
No, he saw how that was for my mom
a few years back and he said he
didn’t want to be kept alive if there
wasn’t a good chance he’d survive.
Provider:
I’m going to give you some time to be
with your dad and process what’s
happening, I know it’s a lot to take
in. But, when I come back, I’d like to
talk to you about our goals for your
dad and how we can make sure
his wishes are honored. That way,
we’ll all be on the same page going
forward, OK?
21
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 21
09/01/15 8:06 AM
SEARL
W W W.A ACNA DVA NCE D CRIT ICA LCA RE .COM
Acknowledgments
I thank Marian Grant, DNP, for support and
guidance in preparing this article.
18.
19.
REFERENCES
1. National Consensus Project for Quality Palliative Care.
Clinical Practice Guidelines for Quality Palliative Care.
2nd ed. Pittsburgh, PA: National Consensus Project for
Quality Palliative Care; 2009.
2. Aslakson RA, Wyskiel R, Thornton I, et al. Nurse-perceived barriers to communication regarding prognosis
and optimal end-of-life care for surgical ICU patients: a
qualitative exploration. J Palliat Med. 2012;15(8):910–
915. doi: 10.1089/jpm.2011.0481.
3. Higgins PC, Prigerson HG. Caregiver evaluation of the
quality of end-of-life care (CEQUEL) scale: the caregiver’s perception of patient care near death. PLoS ONE.
2013;8(6). doi:10.1371/journal.pone.0066066.
4. SUPPORT Principal Investigators. A controlled trial to
improve care for seriously ill hospitalized patients: the
study to understand prognoses and preferences for
outcomes and risks of treatments (SUPPORT). JAMA.
1995;274(20):1591–1598.
5. DeCato TW, Engelberg RA, Downey L, et al. Hospital variation and temporal trends in palliative and end-of-life
care in the ICU. Crit Care Med. 2013;41(6):1405–1411.
doi:10.1097/CCM.0b013e318287f289.
6. Hancock K, Clayton JM, Parker SM. Truth-telling in discussing prognosis in advanced life-limiting illnesses: a
systematic review. Palliat Med. 2007;21:507–517.
doi:10.1177/0269216307080823.
7. Mack JW, Smith TJ. Reasons why physicians do not
have discussions about poor prognosis, why it matters,
and what can be improved. J Clin Oncol. 2012;30:2715–
2717. doi:10.1200/JCO.2012.42.4564.
8. Borowske D. Straddling the fence: ICU nurses advocating for hospice care. Crit Care Nurs Clin N Am.
2010;24:105–116. doi:10.1016/j.ccell.2012.01.006.
9. Evans LR, Boyd EA, Malvar G, et al. Surrogate decisionmakers’ perspectives on discussing prognosis in the
face of uncertainty. Am J Respir Crit Care Med.
2009;179:48–53. doi:10.1164/rccm.200806-969OC.
10. Gillick MR. Decision making near life’s end: a prescription for change. J Palliat Med. 2009;12(2):121–125.
doi:10.1089/jpm.2008.0240.
11. Lind R, Lorem GF, Nortvedt P, Hevrøy O. Family members’ experiences of “wait and see” as a communication strategy in end-of-life decisions. Intensive Care
Med. 2011;37:1143–1150. doi:10.1007/s00134-001-2253-x.
12. Nelson JE, Angus DC, Weissfeld LA, et al. End-of-life care
for the critically ill: a national intensive care unit survey.
Crit Care Med. 2006;34(10):2547–2553. doi:10.1097/01.
CCM.0000239233.63425.1D.
13. Aslakson RA, Pronovost PJ. Health care quality in endof-life care: promoting palliative care in the intensive
care unit. Anesthesiol Clin. 2011;29:111–122. doi:10.1016/
j.anclin.2010.11.001.
14. Field MJ. Approaching death: improving care at the end
of life—a report from the Institute of Medicine. Health
Serv Res. 1998;33(1):1–3.
15. O’Mahony S, McHenry J, Blank AE, et al. Preliminary
report of the integration of a palliative care team into
an intensive care unit. BMC Palliat Care. 2010;24(2):154–
165. doi:10.117/0269216309346540.
16. Kelley MA. Predictive scoring systems in the intensive
care unit. UpToDate. Waltham, MA: UpToDate; 2014.
17. Breslow MJ, Badawi O. Severity scoring in the critically ill
Part 1—Interpretation and accuracy of outcome prediction
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
scoring systems. Chest. 2012;141(1):245–252. doi:10.1378/
chest.11-0330.
Daly BJ, Souglas SL, O’Toole E, et al. Effectiveness trial
of an intensive communication structure for families of
long-stay ICU patients. Chest. 2010;138(6):1340–1348.
Curtis JR, Vincent JL. Ethics and end-of-life care for
adults in the intensive care unit. Lancet. 2010;375:1347–
1353. doi:10.1016/S01406736(10)60143-2.
Puntillo K, Nelson JE, Weissman D, et al. Palliative care
in the ICU: relief of pain, dyspnea, and thirst—A report
from the IPAL-ICU Advisory Board. Intensive Care Med.
2014;40:235–248. doi:10.1007/s00134-013-3153-z.
Vainiola T, Pettilä V, Rione RP, Räsänen P, Rissanen AM,
Sintonen H. Comparison of two utility instruments, the
EQ-5D and the 15D, in the critical care setting. Intensive
Care Med. 2010;36:2090–2093. doi:10.1007/s00134-0101979-1.
Ho LA, Engelberg RA, Curtis JR, et al. Comparing clinician ratings of the quality of palliative care in the intensive care unit. Crit Care Med. 2011;39(5):975–983.
doi:10.1097/CCM.0b13e31820a91db.
Flaatten H. Mental and physical disorders after ICU discharge. Curr Opin Crit Care. 2010;16:510–515. doi:10.1097/
MCC.0b013e32833cc90b.
Unroe M, Kahn JM, Carson SS, et al. One-year trajectories
of care and resource utilization for recipients of prolonged
mechanical ventilation. Ann Intern Med. 2010;153:167–
175. doi:10.7326/0003-4819-153-3-201008030-00007.
Pandharipande PP, Girard TD, Jackson JC. Long-term
cognitive impairment after critical illness. N Engl J Med.
2013;369(14):1306–1306 doi:10.1056/NEJMoa1301372.
Breslow MJ, Badawi O. Severity scoring in the critically ill.
Part 2—maximizing value from outcome prediction scoring systems. Chest. 2012;141(2):518–527. doi:10.1378/
chest.11-0331.
Douglas SL, Daly BJ, Lipson AR. Neglect of quality-of-life
considerations in intensive care unit family meetings for
long-stay intensive care unit patients. Crit Care Med.
2012;40(2):461–467. doi:10.1097/CCM.0b013e318232d8c4.
Vincent JL, Moreno R. Clinical review: scoring systems in
the critically ill. Crit Care. 2010;14(207). doi:10.1186/cc8204.
Keegan MT, Gijic O, Afessa B. Severity of illness scoring
systems in the intensive care unit. Crit Care Med.
2011;39(1):163–169. doi:10.1097/CCM.)b)13e3181f96f81.
Knaus WA, Zimmerman JE, Wagner DP, Draper EA,
Lawrence DE. APACHE—Acute Physiology and Chronic
Health Evaluation: a physiologically based classification system. Crit Care Med. 1981;9(8):591–597.
Zimmerman JE, Kramer AA, McNair DS, Malila FM. Acute
Physiology and Chronic Health Evaluation (APACHE) IV:
hospital mortality assessment for today’s critically ill
patients. Crit Care Med. 2006;34(5):1297–1310.
Le Gall JR, Loirat P, Alperovitch A, et al. A simplified
acute physiology score for ICU patients. Crit Care Med.
1984;12(11):975–977.
Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute
Physiology Score (SAPSII) based on a European/North American multicenter study. JAMA. 1993;270(24):2957–2963.
Lemeshow S, Teres D, Pastides H, Avrunin JS, Steingrub
JS. A method for predicting survival and mortality of
ICU patients using objectively derived weights. Crit
Care Med. 1985;13(7):519–525.
Castella X, Gilabert J, Torner F, Torres C. Mortality prediction models in intensive care: Acute Physiology and
Chronic Health Evaluation II and Mortality Prediction
Model compared. Crit Care Med. 1991;19(2):191–197.
Tunnell RD, Millar BW, Smith GB. The effect of lead time
bias on severity of illness scoring, mortality prediction
and standardized mortality ratio in intensive care—a
pilot study. Anaesthesia. 1998;53(11):1045–1053.
22
Copyright © 2015 American Association of Critical-Care Nurses. Unauthorized reproduction of this article is prohibited.
NCI-D-14-00031.indd 22
09/01/15 8:06 AM