Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Critical illness and intensive care I

Criteria for ICU admission Levels of care defined by the Department of Health

and severity of illness Level Description

scoring 0 Patients whose needs can be met through normal


ward care in an acute hospital
Claire Williams 1 Patients at risk of their condition deteriorating, or
those recently relocated from higher levels of care
Daniel Wheeler
whose needs can be met on an acute ward with
additional advice and support from the critical
care team
Abstract 2 Patients requiring more detailed observation or
In the UK critical care is a limited resource so must be targeted to those intervention, including support for a single failing
patients who are most likely to benefit. Timely recognition and treatment organ system or postoperative care, and those
of the critically ill patient may avoid the need for critical care admis- stepping down from higher levels of care
sion and improve outcome. There is no comprehensive guide to when 3 Patients requiring advanced respiratory support
patients need critical care admission. The support of single-organ failure alone or basic respiratory support together with
can generally be managed in high-dependency care, whereas respiratory support of at least two organ systems. This level
failure or more than two-organ failure requires ICU admission. Track and includes all complex patients requiring support
trigger systems and critical care outreach teams have been introduced for multi-organ failure
following Department of Health advice to identify sick ward patients and
From UK Department of Health.1
ensure they are treated in a timely fashion to avert deterioration into
organ failure. So far there is no conclusive evidence that they achieve
Table 1
these aims. Several severity of illness scores are used in critical care
medicine, which have been developed to compare the performance of
different units, and as a research and audit tool. Such tools predict
outcomes of populations of critically ill patients rather than individuals. In response to this, in 2002 the Intensive Care Society expanded
Choosing which patient to admit still therefore relies on clinicians’ judge- on these levels of care with clearer illustrations of what might
ment in addition to assessment of acute physiological derangement and be appropriate at each level (Table 2).2 Referral on to the next
chronic health status. step must be taken when it is recognized a patient’s level of
­physiological derangement cannot be monitored or managed in
Keywords APACHE; critical care; early warning scores; ICNARC; intensive their current clinical area. It is this that forms a large part of the
care; outreach; scoring systems criteria for admission to critical care.

Timing of admission
In certain surgical patients admission to critical care is a planned
Criteria for admission
event, owing to either patient, surgical or organizational fac-
Levels of care tors (Table 3). It may also become apparent during surgery that
Critical care is a limited resource; the UK Department of Health patients will require more support than originally planned, for
critical care bed count in July 2008 found there were 3498 criti- example because of excessive bleeding or aspiration of gastric
cal care beds, 1970 in ICUs and 1528 in high-dependency units contents. Surgery is also being undertaken on an increasingly
(HDUs). This equates to 8.6 critical care beds per 100,000 head elderly population with substantial co-morbidities. It is the role
of population whereas the USA has 30.5. The UK Department of of the surgeon and anaesthetist to identify the extent to which
Health published ‘Comprehensive critical care: a review of adult the patient’s physiological status can be optimized before surgery
critical care services’ in 2000 in response to this growing aware- to improve outcome.
ness of the lack of critical care beds.1 Levels of care were stated Surgery increases the body’s metabolic needs; these are usu-
in this report; level 2 care is usually delivered in HDUs and level ally met by increasing tissue oxygen delivery. Patients must
3 care in ICUs (Table 1). be assessed to identify those at increased risk of postoperative
hypoxaemia or organ hypoperfusion to allow optimization,
avoiding organ dysfunction or failure.
Claire Williams MA BMBCh FRCA is a Specialist Registrar in Anaesthesia at Oxygen supply to the heart depends on adequate time in dias-
Cambridge University Hospitals NHS Foundation Trust, Addenbrooke’s tole to allow flow within the coronary arteries. Therefore, hypox-
Hospital, Cambridge, UK. Conflicts of interest: none declared. aemia, hypotension and tachycardia decrease myocardial oxygen
supply and increase demand causing ischaemia, especially in
Daniel Wheeler DM PhD MRCP FRCA is a lecturer in Anaesthesia at the patients with pre-existing coronary artery disease. If oxygen
University of Cambridge and an Honorary Consultant Anaesthetist demand remains unmet, ischaemia may progress to infarction,
at Addenbrooke’s Hospital Cambridge. Conflicts of interest: none which carries a perioperative mortality rate of 50%. Hypoten-
declared. sion is defined as inadequate perfusion of vital organs; it may

SURGERY 27:5 201 © 2009 Elsevier Ltd. All rights reserved.


Critical illness and intensive care I

Levels of critical care for adult patients defined by the Intensive Care Society

System Level 0–1 Level 2 Level 3

General Patient recently discharged from Patients needing single-organ system Patients needing advanced respiratory
a higher level of care monitoring and support monitoring and support
Patients in need of additional Patients needing preoperative Patients needing monitoring and support
monitoring, clinical input or optimization for two or more organ systems
advice.
Patients requiring critical care Patients needing extended Patients with chronic impairment of
outreach service support postoperative care one or more organ systems sufficient
to restrict daily activities (co-morbidity)
and who require support for an acute
reversible failure of another organ
system
Patients requiring staff with Patients needing a greater degree of
special expertise and/or observation and monitoring
additional facilities for at least
one aspect of critical care
delivered in a general ward
environment
Patients moving to step-down care
Patients with major uncorrected
physiological abnormalities
Respiratory Supplementary oxygen therapy Needing more than 50% inspired Respiratory failure from any cause that
oxygen. requires invasive, positive-pressure
mechanical ventilatory support
Patients with long-term Within 24 hours of tracheostomy Bi-level positive airway pressure via any
tracheostomy who do not require insertion form of tracheal tube
respiratory support
Requiring non-invasive ventilation or Extracorporeal respiratory support
continuous positive airway pressure
Requiring physiotherapy or suctioning
at least every 2 hours
Respiratory rate  >  40 breaths/minute
or  >  30 breaths/minute for  >  6 hours
Cardiovascular Observations required at least Unstable, requiring continuous ECG Continuous intravenous vasoactive drugs
4 hourly and invasive pressure monitoring
Haemodynamic instability due to
hypovolaemia/haemorrhage/sepsis
Requiring single infusion of vasoactive
drug with appropriate monitoring
Heart rate  >  120 beats/minute
Hypotension, e.g. systolic blood
pressure  <  80  mm Hg for  >  1 hour
CNS CNS depression sufficient to prejudice Continuous intravenous medication to
airway and protective reflexes control seizures and supplementary
oxygen/airway monitoring
Invasive neurological monitoring.
Glasgow Coma Score  <  10 and at risk
of acute deterioration
Renal Renal replacement therapy (stable Acute impairment of renal, electrolyte Renal in addition to another organ
chronic renal failure) or metabolic function system failure

From the Intensive Care Society.2

Table 2

SURGERY 27:5 202 © 2009 Elsevier Ltd. All rights reserved.


Critical illness and intensive care I

As well as optimizing patients before operation it is essential


Factors affecting critical care admission to identify the early warning signs of impending deterioration
and to intervene promptly. The National Confidential Enquiry
Factor Examples into Perioperative Deaths found that most perioperative deaths
occur more than 5 days after surgery. It has emphasized the need
Physiological reserve Proxies for this include age for guidelines to determine which patients should be referred for
and chronic health status critical care review.
Abnormal physiological values The most important are
probably respiratory rate, heart Early warning scores
rate, oxygen saturation (SaO2) Track and trigger (TT) systems have been introduced to ­ identify
and level of consciousness patients at risk of deterioration, and to ensure that they are
Underlying condition/surgery Malignancy reviewed and treated by the appropriate outreach team or clinician.
Cardiorespiratory dysfunction Most employ the recording of standard physiological parameters
Complex surgery which trigger a response if they lie outside a pre-determined range.
Interventions or monitoring Physiological monitoring
In the UK many of these systems are based on an early warning
required Regional analgesia
score (EWS) (Figure 1, Table 4), whereas Australia uses the medi-
Body cavity drainage
cal emergency team (MET) model. A systematic review found there
to be a variety of systems in use, many with little evidence for their
From the Intensive Care Society.2
validity, having low sensitivities and positive predictive values
but with acceptable specificities and negative predictive values.3
Table 3
Lowering the thresholds for a response could improve sensitivity
but at increased cost, greater manpower being required to respond
be caused by hypovolaemia, cardiac failure, arrhythmias, drugs, to the increased number of triggering events.
or spinal or epidural anaesthesia. Persistent hypoperfusion of A London hospital evaluated its local EWS by scoring 433
vital organs causes irreversible damage, leading to kidney, brain non-obstetric patients on a single-day audit; the 30-day mortality
or gut failure. Adequate perfusion is also required for healing; was 6%. Those who died were significantly older and had longer
­hypotension can be a factor in anastomotic breakdown. hospital stays. Mortality increased with an increasing ­ number

Modified Early Warning Score (MEWS) algorithm

MEWS score

Score ≤ 3 but you are


Score 3 Score 3 Score ≥ 4
instinctively concerned

Inform:
Continue present Inform shift charge and 1 Shift charge
Inform shift change
treatment plan critical care outreach team 2 Critical care outreach team
3 Patient’s medical team

Record observations and MEWS


at least hourly for 4 hours

Establish: Oxygen therapy Consider: Fluid resuscitation


Venous access ABG, biochemistry,
12-lead ECG haematology, septic screen
Blood glucose Urinary catheter
Critical care review
Pain control

Yes No
Reduce frequency of observations MEWS ≤ 3 Continue hourly observations
and MEWS to 2 hourly for 4 hours.
If MEWS remains ≤ 3 reduce frequency
of observations to 4 hourly MEWS ≥ 4

Figure 1

SURGERY 27:5 203 © 2009 Elsevier Ltd. All rights reserved.


Critical illness and intensive care I

Modified Early Warning Score used in Addenbrooke’s Hospital, Cambridge, UK

Variable MEWS score

3 2 1 0 1 2 3

Heart rate ≤40 41–50 51–60 61–90 91–110 111–129 ≥130


(beats/minute)
Respiratory rate ≤6 7–8 9–14 15–20 21–29 ≥30
(breaths/minute)
Systolic blood ≤70 71–80 81–100 101–180 >180
pressure (mm Hg)
AVPU/GCS Unresponsive Pain Voice Alert or Confused or 9–13 ≤8
GCS 15/15 GCS 14/15
Urine output <10  ml 0.5  ml/kg in <1  ml/kg/hour Has not passed Has not passed >400  ml/hour or
(catheterized) 2 hours or urine for 6 urine for 12 has not passed
30  ml/hour hours (not hours (not urine for  >  12 hours
catheterized) catheterized) (not catheterized)
Temperature (°C) <35 35–38.4 38.5–39 >39

AVPU, Alert, Voice, Pain, Unresponsive; GCS, Glasgow Coma Score; MEWS, Modified Early Warning Score.

Table 4

of abnormal physiological variables. Level of consciousness,


Severity of illness scoring
heart rate, age, systolic blood pressure and respiratory rate were
important predictive variables. There is no current evidence that Scoring systems have been developed to assess single disease
TT systems reduce mortality; they should be used to complement states and global health status, as a research tool and as a
clinical judgement, not in isolation. ­mechanism for assessing the performance of critical care units.
These scoring systems can be used to generate a standardized
Critical care outreach mortality ratio, the ratio of actual to predicted mortality for each
Another recommendation of ‘Comprehensive critical care’ was unit. This takes into account factors such as case mix (differ-
the introduction of critical care outreach services (CCOS). Their ences in the pre-morbid health of patients) to allow an unbiased
objectives are to avert or ensure timely critical care admission, comparison of different units or the same unit over time. Most
to enable prompt discharge from critical care and to share skills of these systems were developed in North America; it is unclear
with ward staff. This document did not provide any models for how they apply to other populations. Most systems evaluate the
CCOS and they were introduced without any prospective evalu- patient’s condition on arrival in the ICU but do not take into
ation. A multicentre analysis of CCOS participating in the Case account treatment received in theatre, the emergency department
Mix Programme had interesting findings.4 The introduction of or on the ward beforehand. Lead-time bias may affect the calcu-
CCOS had no significant effect on the proportion of all admissions lated severity score and apparent unit performance.
coming from the ward. CCOS significantly reduced the amount
of ­cardiopulmonary resuscitation (CPR) in the 24 hours preced- Physiological scores
ing admission, admissions out of hours (between 2200 and 0659 Physiological scores are summarized in Table 5.
hours) and the mean Intensive Care National Audit and Research
Centre (ICNARC) physiology score. The additional use of TT American Society of Anesthesiologists (ASA) grade: this system
warning systems reduced rates of CPR before admission. There has been in use since 1963. It grades preoperative patients from I
was no effect on mortality rates of patients admitted, mortality to V according to the degree of systemic illness, with the addition
after discharge from critical care or readmission rates. This paper of the tag of emergency when indicated (Table 6). Although it is
did not show any clear benefit to the introduction of CCOS. Stud- subjective, it is simple and it has been shown repeatedly that risk
ies of METs in Australia have shown a benefit to patients under- of mortality increases with increasing ASA grade; 50% of surgical
going major surgery; there were fewer adverse events, decreased deaths are in patients with an ASA score of IV or V.
emergency ICU admissions and fewer postoperative deaths.
The Cochrane Collaboration has reviewed both outreach Acute Physiology And Chronic Health Evaluation (APACHE):
services and EWS.5 It found only two randomized controlled the APACHE model was first developed in 1981 with data from
trials appropriate for inclusion in its review. Evidence to sup- two American ICUs; it included 34 physiological variables in
port CCOS and EWS in reducing hospital mortality, unplanned addition to an assessment of chronic ill health. The simplified
ICU admission, readmission and length of hospital stay was APACHE II included data from 13 American hospitals, relying
­inconclusive. on only 12 variables to create the acute physiology score (APS).

SURGERY 27:5 204 © 2009 Elsevier Ltd. All rights reserved.


Critical illness and intensive care I

Summary of physiological scoring systems

Scoring System Date published Origin of patient database Variables in APS Time of data collection Exclusions

APACHE II 1985 USA 12 First 24 hours <16 years


APACHE IV 2006 USA 16 First 24 hours <16 years
SAPS II 1994 USA/Europe 13 First 24 hours <18 years, coronary care,
burns and cardiac surgery
SAPS III 2005 Global 10 Admission +/− 1 hour <16 years
MPM0-II 1993 USA/Europe 15 First hour <18 years, coronary care,
burns and cardiac surgery
MPM0-III 2007 USA/Europe 16 First hour <18 years, coronary care,
burns and cardiac surgery
POSSUM 1991 UK 12 Any Non-operative patients
ICNARC 2007 UK 12 First 24 hours None

APS, acute physiology score; APACHE, Acute Physiology And Chronic Health Evaluation; ICNARC, Intensive Care National Audit and Research Centre; MPM, Mortal-
ity Prediction Model; POSSUM, Physiology and Operative Severity Score for EnUmeration of Mortality and morbidity; SAPS, Simplified Acute Physiology Score.

Table 5

The further addition of the chronic health evaluation, age and SAPS III which also includes patient characteristics before admis-
urgency of admission undergo logistic regression to give the total sion and circumstances of admission. The dataset came from an
APACHE score, a probability of mortality. APACHE III was devel- international group of patients; the predictive value varies glob-
oped with data from an expanded set of 40 American hospitals ally, with performance being best in North Europe.
and an 18-value APS. The statistical models used for this system
have not been in the public domain so it has not been as widely Mortality Prediction Model (MPM) was first developed in the
used and validated as the first two models. Recently APACHE late 1980s and included data on pathology only. The MPM at
IV was published in response to the decreasing accuracy of the ICU admission version 2 (MPM0-II) was developed from data
­previous versions.6 This is believed to be due to changes in out- ­collected from an international group of patients. The data are
come with time given advances in medical care and improve- collected from the first hour of a patient’s admission to the ICU.
ments in end-of-life decision making. APACHE II is still in most It was felt that this model overpredicted mortality so MPM-III
common use owing to its simplicity and numerous validation was developed from a dataset of 124,885 patients.7 This model
studies, but the APACHE authors have requested that APACHE II uses 16 variables recorded on admission to predict mortality.
no longer be used, except in patients with severe septis, so that One advantage of this system is that it can be repeated at 24,
the newer models can be validated. 48 and 72 hours to track changes during a critical care stay.
Again it has been refined over the years to improve calibration.7
Simplified Acute Physiology Score (SAPS) employed 14 of the The most recent version (MPM0-III) is best employed on general
original 34 variables of APACHE. The most recent iteration is ICUs as it excludes children, patients with burns and cardiac
patients.

Physiology and Operative Severity Score for EnUmeration of


American Society of Anesthesiologists grades Mortality and morbidity (POSSUM): unlike the other systems
discussed this was developed for use in surgical patients. Simi-
ASA grade Definition Mortality
lar to the others, it was developed for audit and not individual
I Normal healthy individual 0.05 patient prognosis. It employs a combination of physiological
II Mild systemic disease that does not 0.4 derangement and an operative severity score (OSS). This OSS
limit activity means that the system is applicable only to operative surgical
III Severe systemic disease that limits 4.5 patients, excluding the one-third of patients who are managed
activity but is not incapacitating non-operatively.
IV Incapacitating systemic disease that 25
constantly threatens life ICNARC: APACHE II was developed from US data and the UK
V Moribund, not expected to survive 50 APACHE II study validated this model in UK patients. This led
24 hours to the development of the Case Mix Programme as an ongoing
national critical care audit and the ICNARC model.8 ICNARC
ASA, American Society of Anesthesiologists. differs from other systems by employing a diagnostic cate-
gory to account for the fact that a patient with diabetic keto-
Table 6 acidosis will have a lower risk of mortality than a patient with ­­

SURGERY 27:5 205 © 2009 Elsevier Ltd. All rights reserved.


Critical illness and intensive care I

intra-­abdominal sepsis, given the same degree of physiological were developed from a patient database, MODS was developed
derangement. ICNARC allows a less biased appraisal of ICUs from published clinical studies.
with different admission case mixes. It also includes patients
who have been excluded from other models, such as children
Conclusion
and burns patients. The resulting model is therefore applicable
to a larger range of UK patients and is less biased when compar- This article has focused on who to admit to the ICU and when.
ing different ICUs. We must also not lose sight of those patients for whom admis-
sion is inappropriate. Scoring systems may guide us but clinical
Anatomical scores judgement and sensitive communication remain core skills when
Disease and organ-specific systems include the Glasgow Coma dealing with end-of-life events and decisions. ◆
Score (GCS), Ranson pancreatitis score and Child–Pugh liver fail-
ure classification. The use of these is not limited to critical care
and they can be useful in documenting disease severity. The GCS References
was originally developed for use in trauma patients and traumatic 1 Department of Health. Comprehensive critical care: a review of adult
brain injury, but its use is now far more widespread. In terms of critical care services. London: Department of Health, 2000.
critical care, it should be used only in specialist neurointensive 2 Intensive Care Society. Levels of critical care for adult patients.
care units and not to describe patients who are unconscious for London: Intensive Care Society, 2002.
other reasons. 3 Gao H, McDonnell A, Harrison DA, et al. Systematic review and
evaluation of physiological track and trigger warning systems for
Organ dysfunction scores identifying at-risk patients on the ward. Intensive Care Med 2007;
Most of the systems described above are static; they assess 33: 667–79.
patients only on admission or within the first 24 hours of admis- 4 Gao H, Harrison DA, Parry GJ, Daly K, Subbe CP, Rowan K. The impact
sion to critical care. In many cases organ and physiological func- of the introduction of critical care outreach services in England: a
tion fluctuate following admission, so dynamic scoring might multicentre interrupted time-series analysis. Crit Care 2007; 11: R113.
therefore be more appropriate. 5 McGaughey J, Alderdice F, Fowler R, Kapila A, Mayhew A, Moutray M.
Outreach and Early Warning Systems (EWS) for the prevention of
Sepsis-related Organ Failure Score (SOFA) was published in intensive care admission and death of critically ill adult patients
1994 by the European Society of Intensive Care Medicine fol- on general hospital wards. Cochrane Database Syst Rev 2007; (3)
lowing a consensus conference. It involves daily scoring of six CD005529.
organ systems to allow an organ-by-organ, total or worst-during- 6 Zimmerman JE, Kramer AA, McNair DS, Malila FM. Acute Physiology and
admission score. Although it was developed to be independent of Chronic Health Evaluation (APACHE) IV: hospital mortality assessment
treatment, the cardiovascular score includes use of a vasoactive for today’s critically ill patients. Crit Care Med 2006; 34: 1297–310.
drug. This system has been validated in both a retrospective and 7 Higgins TL, Teres D, Copes WS, Nathanson BH, Stark M, Kramer AA.
prospective study. Assessing contemporary intensive care unit outcome: an updated
Mortality Probability Admission Model (MPM0-III). Crit Care Med
Multiple Organ Dysfunction Score (MODS) scores six organ sys- 2007; 35: 827–35.
tems on a scale from 0 to 4, where 0 is essentially normal func- 8 Harrison DA, Rowan KM. Outcome prediction in critical care: the
tion and the maximum total score is 24. MODS correlates with ICNARC model. Curr Opin Crit Care 2008; 14: 506–12.
ICU outcome, and can be used to assess prognosis on admission 9 Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ.
and as an outcome measure, tracking organ dysfunction as it Multiple organ dysfunction score: a reliable descriptor of a complex
varies during ICU admission.9 Unlike other scoring systems that clinical outcome. Crit Care Med 1995; 23: 1638–52.

SURGERY 27:5 206 © 2009 Elsevier Ltd. All rights reserved.

You might also like