From the Departments of *Medicine, Clinical Epidemiology and Anthropology, and the Centre For Health Economics and Policy Analysis, McMaster University, Hamilton, Ont.
The ability of many intensive care unit (ICU) technologies
to prolong life has led to an outcomes-oriented approach to
technology assessment, focusing on morbidity and mortality as
clinically important end points. With advanced life support,
however, the therapeutic goals sometimes shift from extending
life to allowing life to end. The objective of this study was
to understand the purposes for which advanced life support is
withheld, provided, continued or withdrawn in the ICU.
In a 15-bed ICU in a university-affiliated hospital,
the authors observed 25 rounds and 11 family meetings in which
withdrawal or withholding of advanced life support was addressed.
Semi-structured interviews were conducted with 7 intensivists,
5 consultants, 9 ICU nurses, the ICU nutritionist, the hospital
ethicist and 3 pastoral services representatives, to discuss
patients about whom life support decisions were made and to
discuss life-support practices in general. Interview transcripts
and field notes were analysed inductively to identify and corroborate
emerging themes; data were coded following modified grounded
theory techniques. Triangulation methods included corroboration
among multiple sources of data, multidisciplinary team consensus,
sharing of results with participants and theory triangulation.
Although life-support technologies are traditionally
deployed to treat morbidity and delay mortality in ICU patients,
they are also used to orchestrate dying. Advanced life support
can be withheld or withdrawn to help determine prognosis. The
tempo of withdrawal influences the method and timing of death.
Decisions to withhold, provide, continue or withdraw life support
are socially negotiated to synchronize understanding and expectations
among family members and clinicians. In discussions, one discrete
life support technology is sometimes used as an archetype for
the more general concept of technology. At other times, life-support
technologies are discussed collectively to clarify the pursuit
of appropriate goals of care.
The orchestration of death involves process-oriented
as well as outcome-oriented uses of technology. These uses should
be considered in the assessment of life-support technologies
and directives for their appropriate use in the ICU.
Critical care medicine provides 2 major services for seriously ill patients: intense and sometimes invasive diagnosis and monitoring, to allow early recognition and treatment of biomedical problems, and advanced life support, to improve the short, and possibly long-term survival of patients with exigent, life-threatening illness.1 Critical care medicine uses state-of-the-art technology to pursue its mission.
The dramatic ability of many intensive care unit (ICU) technologies to prolong life has led to an outcomes-oriented approach to technology assessment, focusing on morbidity and mortality as clinically important end points.2 In the case of advanced life support, however, the therapeutic goals sometimes shift from extending life to allowing life to end. Recent Canadian health research has shifted from matters of life-support administration to issues in life-support discontinuation.3-7 Concurrent with this trend are qualitative investigations into end-of-life decision-making8-10 and understanding the ICU as a social world,11,12 as well as calls to assess the ethical and social influences of biomedical technologies.13,14 However, biomedical, evaluative, ethical and social science studies of life-support technology remain poorly integrated.
These combined disciplinary perspectives can be used to examine
the diverse purposes of life-support technologies as they are
used in practice. "Real" technological purposes can then be
addressed more explicitly in assessment exercises.15,16
objective of our study was to explore the purposes for which
advanced life support is used in the care of critically ill,
dying patients who are unable to make their own decisions.
The descriptive aim of the research, and the social nature of the subject matter, called for a qualitative, naturalistic approach to inquiry.17 The study was conducted over 14 months in the 15-bed closed ICU of St. Joseph’s Hospital, Hamilton, Ont. In 1-week blocks 52 full-time nurses, 25 part-time nurses and 7 intensivists attend the ICU; every 2 months, 4 junior residents rotate through the ICU.
We observed 25 ICU rounds and 11 family meetings during which
withdrawal or withholding of advanced life support was discussed.
Eleven cases of life-support decision-making were observed.
Extensive field notes were recorded. Semi-structured interviews
were also conducted with the 7 intensivists, 5 consultants,
9 nurses, the ICU nutritionist, the hospital ethicist and 3
pastoral services representatives. Interviewees were purposely
selected on the basis of their involvement with patients from
whom life support technology was withheld or withdrawn during
the study period. We were unable to communicate with ICU patients
about whom life support decisions were made because of their
cognitive status or level of sedation. Interviews addressed
patients about whom advanced life support was discussed, as
well as life-support practices and issues in general. Interviews
were audiotaped and transcribed verbatim.
Transcripts of the interviews and field notes were managed using
qualitative research software. Qualitative analysis focused
on interpreting the meaning of participants’ discussions and
actions regarding end-of-life decision-making. Transcripts and
field notes were analysed inductively to identify and corroborate
emerging themes. Data were coded following modified grounded
the coding scheme was revised and refined
over the course of the study. We audiotaped investigator meetings,
and these transcripts became part of the database. Several triangulation
were used to validate key findings: corroboration
among multiple sources of data, multidisciplinary team consensus
(the research team included 1 intensivist, 2 medical anthropologists
and 1 policy analyst) and sharing of results with participants
for veracity of interpretations. Theory triangulation (relating
the findings to those of published studies on similar problems,
concepts and settings) was used to contextualize the findings
and guide later stages of the analysis.
The human subjects protocol for this project was approved by
the St. Joseph’s Hospital and McMaster University Research Ethics
Boards. Participation was voluntary, and informed consent was
obtained from participants. Confidentiality and data security
measures were also reviewed with participants in the consent
Life support is characterized by the use of a panoply of technologies
and myriad decisions that unfold over the course of an ICU stay.
For patients who eventually die, the administration and withdrawal
of life support is particularly complex. Decisions to withdraw
life support barely resemble the popular imagery of "pulling
the plug." Withdrawal is not a decisive event but, rather, an
unfolding process. The metaphor of the "orchestration" of death
describes the process of determining which life-support technologies
come into play, to what ends, when, by whom and for whom.
Life-support technologies are instrumental beyond reducing morbidity
or preventing mortality. In the context of end-of-life decision-making,
clinicians use technology to orchestrate the "best" death possible
for critically ill patients under difficult circumstances. This
goal is concerned less with health outcomes in the traditional
sense than it is with the aesthetic, ethical and social experiences
of those involved in the patient’s care (e.g., significant others,
family members and clinicians). In this context, technologies
might be considered as analogous to orchestral instruments for
expressing values and visions, as well as clinical instruments
for producing health.
Technological silence for prognostication
Patients receiving advanced life support often have unclear
diagnoses and prognoses, particularly early in the course of
their illness. Life support is continued while prognostic uncertainty
is addressed. Clinicians often speak of waiting for patients
to "declare themselves," that is, for clearer prognostic signs
The physician may spend a lot of time talking to families, trying
to get them to understand that this patient is not going to
make it. Sometimes, the family never does give in, but then
the patient declares himself and just starts to deteriorate
anyway, and the physician tells the family, "there is nothing
else we can do." [interview with ICU nurse]
Life-support technology can also be withdrawn for prognostic
purposes or stopped on a trial basis. The ensuing technological
silence creates an opportunity for patients to declare themselves.
Dialysis may be discontinued to determine whether renal failure
will reverse. Mechanical ventilation may be stopped to assess
whether spontaneous breathing can be sustained.
[The patient] was extubated on Friday and did okay over the
evening, but by early Saturday he had tuckered out and required
reintubation, and then Monday he was extubated again. This time
he lasted only two hours. Again, it didn’t appear that he was
ready for extubation. His parameters, weaning-wise, never really
looked that good, but it was sort of … "Well, let’s just see
how he does." [interview with ICU resident]
The tempo of life support and dying
The withholding or withdrawal of life support can be orchestrated
to occur quickly or slowly, changing the tempo of dying.
What appears to be happening really varies among staff persons
as to how people are extubated. Some people are very aggressive
and just take the tube out and see how they do. If they’re going
to make it, they’ll make it. Other people are more conservative
in a sense, using a stepwise approach and slowly bringing down
the pressure and stuff like that. There were three different
ways this person could be extubated, so it varies. [interview
with ICU resident]
The pace at which life-support technologies are withdrawn, and
the sequential order of withdrawal, may be influenced by many
concerns. These include the potential suffering experienced
by the patient, vicarious suffering experienced by others and
the speed of the consequent death.
Some people walk in and yank the endotracheal tube and others
will say "let’s stop the drugs, let’s stop the oxygen." I have
trouble yanking out the endotracheal tube probably because I
think that it increases the chances that the patient is going
to die actively trying to breathe against an obstructed airway.
I don’t think that’s a nice way to die. I find it a little tougher
to do that than to say, "I think if we turn off the drug he’s
not going to last very long." For me, personally, it’s a lot
easier to turn off the drug. I guess it relates to how I see
the patient’s comfort. [interview with intensivist]
Practice variations with respect to terminal weaning or ventilator
withdrawal have long been recognized but only recently publicized.22
The observation that life support is withdrawn sequentially,
rather than all at once, is supported by findings from a retrospective
multicentre US study.23
Other research suggests that physicians
preferentially withdraw forms of life support that are scarce,
expensive, invasive, artificial or emotionally taxing.24
Life-support technologies in the ICU may be applied continuously
(e.g., mechanical ventilation for respiratory failure) or intermittently
(e.g., hemodialysis for renal failure). Interventions may also
be sequenced, such as the alternation of defibrillation with
pharmacology in cardiac resuscitation protocols. The rhythmic
nature of these acts is often overlooked as a feature of life-support
decision-making. Orchestration decisions concern how long, how
often and whether to use a particular technology.
Harmonizing expectations and decisions
Hours or days may pass from the time a patient’s condition irretrievably worsens and the time life support ends. Life support creates an interlude during which people strive to harmonize their understandings, expectations and plans for the patient. Family members and clinicians work, and wait, for the synchronous acceptance of futility and imminent death. The family may need time to overcome denial that the patient is dying, disbelief that treatment options have run out or disagreements among themselves that death is inescapable. People need time to say goodbye.
Mr. A’s son explained that some of [the family’s] concern about
time frames was due to the fact that Mr. A’s remaining sister
would not be able to arrive until Saturday morning (this discussion
took place on Thursday afternoon). Dr. B conveyed that he personally
had "no trouble continuing to support Mr. A until Saturday,
if it is very important that Mr. A’s sister see her brother."
As with many clinicians, family members may have preferences
and feelings about how life support is withdrawn. This period
may also be used for negotiation between the patient’s wishes
(usually hypothesized, rarely known) and the family members’
and clinicians’ personal views about specific technologies,
their uses and likely effects.
Mrs. C and Dr. D arrived at a decision to begin the withdrawal
process by decreasing the amount of oxygen Mr. C was receiving
from 45% to 35%. Mr. C died about 18 hours later – "peacefully,
in his sleep." Dr. D concluded, saying that "nothing had been
really withdrawn." When questioned if the oxygen being reduced
was not something being withdrawn, Dr. D replied that Mrs. C
was comfortable with it being turned down because it seemed
to be "more natural."
Conducting the orchestra versus the instruments
At key points in life-support decision-making, family members
and caregivers relate to life-support technologies collectively
as "technology" itself, rather than as discrete technological
tools with specific therapeutic uses. Technology thus comes
to represent a global approach to achieving the goals of care.
Typically, this happens early in the ICU stay, when there is
a desire to "do everything" and the objective is saving life.
I would say in general the majority of times the family will
say "do everything." There are some families that will say "Oh,
no, do nothing, let’s stop now," but the majority want to push
on initially, and it’s only usually after fairly extensive discussions
that that viewpoint will shift at all, and sometimes it never
does. [interview with consultant]
The orchestration of life support often concerns the number
of technologies in play at once, and whether to add new instruments
to the mix. Life-support technologies initiated early on under
a more optimistic prognosis or aggressive management plan may
be continued, but additional interventions or life-support measures
may be withheld. At issue is the intensity of care overall,
rather than the merits and demerits of specific interventions.
Dr. B stressed that it was his opinion, however, that they should
not "begin new things or add things." If Mr. A were to get worse
between now and Saturday, he would not do anything new to support
him. To do so would be "unkind."
A collective view of technology similarly appears toward the
end of the ICU stay, as an imperative to "stop doing everything."
In this context, the use of technology seems aesthetically or
morally offensive, as dehumanizing or degrading to the body.
The inherent goal of technology has shifted from life-saving
to death-prolonging or pain-inducing. In discussions, erstwhile
useful instruments transform dismissively into "a whole bunch
of machines" and a source of discord.
She says that above all, she wants him to be comfortable. "No
more artificial machines."
When life support
is withdrawn, they usually ask the family to leave the room
while the machines and equipment are removed. Once everything
is gone, and it is just the patient in the bed, the family comes
back until death occurs.
In some discussions about the withdrawal of life support, a
discrete technology (e.g., dialysis or mechanical ventilation)
can act as an archetype or synecdoche for the more general concept
of life-support technology and life-support goals. By working
through a decision about whether to use dialysis, for example,
clinicians and family members can begin to address decisions
regarding other life-support technologies more generally.
In the ICU, the decision to stop dialysis is generally tantamount
to death. And usually what will happen is that we’ll stop dialysis
and the family will say, "Well, continue everything else," and
then a couple of days later they’ll have thought about it a
bit more if the patient hasn’t already died, and say, "Well,
maybe we should start withdrawing other things as well." It’s
just sort of the first step on the path that eventually leads
to withdrawal of active care. [interview with consultant]
When life-support technology is used to orchestrate a death in the ICU, it can perform functions not well appreciated by conventional frameworks for technology evaluation. End-of-life decision-making concerns not only whether to use life-support technology, but also how – the timing, intensity and number of technologies. These decisions are socially negotiated and nuanced for each patient. Goals of technology use in this context include explicit social, aesthetic and ethical objectives as well as clinically instrumental objectives.
We have identified some of these technological functions. Although
life support is commonly understood to be therapeutic, it may
also be withdrawn on a trial basis to gain diagnostic and prognostic
information. Life support may be provided continuously or intermittently.
When it is withdrawn, it can be done abruptly or slowly, in
numerous discretionary sequences, and to a number of clinical,
social and psychological effects on patients, family members
and clinicians. Periods of seemingly futile life support may
create an interlude in which family members can come to terms
with and negotiate the dying process. Some life-support decisions
are about specific instruments, while others are about instrumentation,
or the image of technology, in general.
These findings suggest several implications for practice and
policy. Models of the physician-patient relationship are often
portrayed as models of decision-making. This conflation seems
particularly problematic in the ICU, where patient-clinician
relationships may be new or undeveloped, where emotional tensions
are unparalleled, and where unanticipated situations may eclipse
advance health care planning. Therefore, models of patient-physician
require adaptation to be relevant to decisions
concerning advanced life support in the ICU. A given technological
act (e.g., withdrawal of mechanical ventilation) may have multiple
clinical and social meanings. Decisions may concern technology
in general rather than specific technologies. Many discussions
about styles of withdrawing mechanical ventilation leave key
technologic options implicit. Individual technologies may be
tacitly understood to achieve overall goals. Our findings also
raise questions about the ability of increasingly popular decision
aids such as advance directives to portray the complex purposes,
effects and meaning of life-support technologies as they are
actually used and experienced by patients, their family members
A key issue is implied but not directly addressed in this report.
Patient autonomy notwithstanding, the orchestration metaphor
generated by our analysis alludes to "composers" and "conductors"
who coordinate how technologies play out in the ICU. The influences
of social dynamics, culture and consensus building on perceptions
of the form and function of ICU technology are beyond the scope
of this report,26 
but they represent fruitful areas for investigation.
Future frameworks of life-support withholding, administration,
continuation and withdrawal should accommodate the multiple
functions of technology, including its social and personal impact
on dying and grieving.
We thank the family members, nurses, house staff, physicians
and other health care workers who participated in this study.
We also thank the ICU nurses and coordinators, Dr. Michael Coughlin,
Patricia Upton, RN, Ellen McDonald, RN, Barbara Hill and the
St. Joseph’s Hospital Administration for helping to create the
culture that supported this research.
This article has been peer reviewed.
This study was funded by the Medical Research Council of Canada
and the Father Sean O’Sullivan Research Centre. Dr. Cook is
a Career Scientist of the Ontario Ministry of Health. Dr. Giacomini
is a National Health Research Scholar of Health Canada and a
member of the Centre for Health Economics and Policy Analysis,
which is funded in part by the Ontario Ministry of Health.
Reprint requests to: Dr. Deborah J. Cook, Department of Medicine,
St. Joseph’s Hospital, 50 Charlton Ave. E, Hamilton ON L8N 4A6;
Cook DJ, Sibbald WJ, Vincent JL, Cerra FB, for the Evidence Based Medicine in Critical Care Group. Evidence based critical care medicine: What is it and what can it do for us? Crit Care Med
Sibbald WJ, Inman KJ. Problems in assessing the technology of critical care medicine. Int J Technol Assess Health Care
Wood GG, Martin E. Withholding and withdrawing life sustaining therapy in a Canadian intensive care unit. Can J Anaesth
Keenan SP, Busche KD, Chen LM, McCarthy L, Inman KJ, Sibbald WJ. A retrospective review of a large cohort of patients undergoing the process of withholding or withdrawal of life support. Crit Care Med
Keenan SP, Busche KD, Chen LM, Esmail R, Inman KJ, Sibbald WJ, for the Southwestern Ontario Critical Care Research Network. Withdrawal and withholding of life support in the intensive care unit: a comparison of teaching and community hospitals. Crit Care Med
Cook DJ, Guyatt GH, Jaeschke R, Reeve J, Spanier A, King D, et al. Determinants in Canadian health care workers of the decision to withdraw life support from the critically ill. JAMA
Walter S, Cook DJ, Guyatt GH, Spanier A, Jaeschke R, Todd T, et al. Confidence in life support decisions in the ICU: a survey of health care workers. Crit Care Med
Snider G. Withholding and withdrawing life-sustaining therapy: All systems are not yet "go." Am Rev Respir Crit Care Med
Ventres W, Nichter M, Reed R, Frankel R. Limitation of medical care: an ethnographic analysis. J Clin Ethics
Singer PA, Martin DK, Kelner M. Quality end-of-life care: patients’ perspectives. JAMA
Zussman R. Intensive care.
Chicago: University of Chicago Press; 1992.
Chambliss DF. Death as an organizational act. In: Beyond caring: hospitals, nurses, and the social organization of ethics.
Chicago: University of Chicago Press; 1996. p. 150-79.
Good MJ. Cultural studies of biomedicine: an agenda for research. Soc Sci Med
Gjengedal E. The ethical impact of advanced biomedical technology: on means and ends in high-technology medicine. Scand J Caring Sci
Ten Have HAMJ. Medical technology assessment and ethics: ambivalent relations. Hastings Cent Rep
Giacomini M. The "which" hunt: assembling health technologies for assessment and rationing. J Health Polit Policy Law.
Lincoln YS, Guba EG. Naturalistic inquiry.
London: Sage Publications; 1985.
Strauss A, Corbin J. Open coding. In: Basics of qualitative research: grounded theory procedures and techniques.
London: Sage Publications; 1990. p. 61-74.
Strauss A, Corbin J. Axial coding. In: Basics of qualitative research: grounded theory procedures and techniques.
London: Sage Publications; 1990. p. 96-115.
Strauss A, Corbin J. Selective coding. In: Basics of qualitative research: grounded theory procedures and techniques.
London: Sage Publications; 1990. p. 116-42.
Lincoln YS, Guba EG. Establishing trustworthiness. In: Naturalistic inquiry.
London: Sage Publications; 1985. p. 289-331.
Faber-Langendoen K. The clinical management of dying patients receiving mechanical ventilation: a survey of physician practice. Chest
Faber-Langendoen K. A multi-institutional study of care given to patients dying in hospitals. Arch Intern Med
Asch DA, Christakis NA. Why do physicians prefer to withdraw some forms of life support over others? Intrinsic attributes of life-sustaining treatments are associated with physicians preferences. Med Care
Emanuel EJ, Emanuel LL. Four models of the physician-patient relationship. JAMA
Rosenfield PL. The potential of transdisciplinary research for sustaining and extending linkages between the health and social sciences. Soc Sci Med
Yaguchi, R. D. Truog, J. R. Curtis, J. M. Luce, M. M. Levy, C. Melot, and J.-L. VincentInternational Differences in End-of-Life Attitudes in the Intensive Care Unit: Results of a Survey
Archives of Internal Medicine, September 26, 2005; 165(17): 1970 – 1975.
C. R. Levy, E. W. Ely, K. Payne, R. A. Engelberg, D. L. Patrick, and J. R. Curtis
Quality of Dying and Death in Two Medical ICUs: Perceptions of Family and Clinicians
Chest, May 1, 2005; 127(5): 1775 – 1783.
R. I. Hall, G. M. Rocker, and D. MurraySimple changes can improve conduct of end-of-life care in the intensive care unit: [Des changements simples peuvent ameliorer les soins aux mourants a l’unite des soins intensifs]
Can J Anesth, June 1, 2004; 51(6): 631 – 636.
G. M. Rocker, D. K. Heyland, D. J. Cook, P. M. Dodek, D. J. Kutsogiannis, and C. J. O’CallaghanMost critically ill patients are perceived to die in comfort during withdrawal of life support: a Canadian multicentre study: [Les grands malades meurent sans souffrance pendant le retrait du maintien des fonctions vitales : une etude canadienne]
Can J Anesth, June 1, 2004; 51(6): 623 – 630.
S. D. Shemie, C. Doig, and P. BelitskyAdvancing toward a modern death: the path from severe brain injury to neurological determination of death
Can. Med. Assoc. J., April 15, 2003; 168(8): 993 – 995.
T. J. Prendergast and K. A. PuntilloWithdrawal of Life Support: Intensive Caring at the End of Life
JAMA, December 4, 2002; 288(21): 2732 – 2740.
M. CusimanoThe journey through the ICU
Can. Med. Assoc. J., March 1, 2000; 162(5): 633 – 634.
D. J. Cook, M. O. Meade, and A. G. PerryQualitative Studies on the Patient’s Experience of Weaning From Mechanical Ventilation
Chest, December 1, 2001; 120(90060): 469S – 473.
R. I. Hall and G. M. RockerEnd-of-Life Care in the ICU : Treatments Provided When Life Support Was or Was Not Withdrawn
Chest, November 1, 2000; 118(5): 1424 – 1430.