Davis N, Pohlman A, Gehlbach B, Kress JP, McAtee J, Herlitz J, Hall J. Improving the Process of Informed Consent in the Critically Ill. JAMA. 2003;289(15):1963-1968. doi:10.1001/jama.289.15.1963
Author Affiliations: Section of Pulmonary and Critical Care Medicine, the Pritzker School of Medicine, and the Office of Risk Management, University of Chicago, Chicago, Ill.
Caring for the Critically Ill Patient Section Editor: Deborah J. Cook, MD, Consulting Editor, JAMA.
Context Invasive procedures are often performed emergently in the intensive
care unit (ICU), and patients or their proxies may not be available to provide
informed consent. Little is known about the effectiveness of intensivists
in obtaining informed consent.
Objectives To describe the nature of informed consent in the ICU and to determine
if simple interventions could enhance the process.
Design, Setting, and Patients Prospective study of 2 cohorts of consecutively admitted patients (N
= 270) in a 16-bed ICU at a university hospital. All patients admitted to
the ICU during the baseline period from November 1, 2001, to December 31,
2001, and during the intervention period from March 1, 2002, to April 30,
2002, were included.
Intervention A hospital-approved universal consent form for 8 commonly performed
procedures (arterial catheter, central venous catheter, pulmonary artery catheter,
or peripherally inserted central catheter placement; lumbar puncture; thoracentesis;
paracentesis; and intubation/mechanical ventilation) was administered to patients
or proxies. Handouts describing each procedure were available in the ICU waiting
area. Physicians and nurses were introduced to the universal consent form
during orientation to the ICU.
Main Outcome Measures Incidence of informed consent for invasive procedures at baseline and
after intervention; whether the patient or proxy provided informed consent;
and understanding by the consenter of the procedure as determined by the responses
on a questionnaire.
Results Fifty-three percent of procedures (155/292) were performed after consent
had been obtained during the baseline period compared with 90% (308/340) during
the intervention period (absolute difference, 37.4%; P<.001).
During baseline, the majority (71.6%; 111/155) of consents were provided by
proxies. This was also the case during the intervention period in which 65.6%
(202/308) of consents were provided by proxies (absolute difference, 6.0%; P = .23). Comprehension by consenters of indications for
and risks of the procedures was high and not different between the 2 periods
(P = .75).
Conclusions Invasive procedures are frequent in the ICU and consent for them is
often obtained by proxy. Providing a universal consent form to patients, proxies,
and health care clinicians significantly increased the frequency with which
consent was obtained without compromising comprehension of the process by
Informed consent is the process by which a person authorizes medical
treatment after discussion with clinicians regarding the nature, indications,
benefits, and risks of the treatment.1 In the
past, physicians did not routinely seek permission from patients and/or proxies
to provide medical treatment, even when such treatments involved invasive
procedures with significant risks.2 However,
in current practice, such physician sovereignty has been replaced with an
emphasis on patient autonomy.3 At present,
physicians are obliged to expect and encourage patient participation in decisions
regarding care. The process of obtaining informed consent can improve patient
satisfaction and health outcomes as well as engender trust in the physician
and compliance with treatment recommendations.1,4,5
Patients receiving care in the intensive care unit (ICU) represent a
highly vulnerable population with regard to informed consent. While many invasive
procedures often accompany the provision of life support in the ICU and are
frequently required for diagnostic or therapeutic purposes, critically ill
patients are often not capable of participating in the consent process. This
is the result of the common incidence of delirium complicating critical illness,
related to both underlying illnesses and the use of sedatives and analgesics
to treat anxiety and pain.6- 9 Because
of this, proxies are often required to provide health decisions for impaired
patients.10 Yet the pace of critical illness
is often rapid and unpredictable, and situations may arise that do not permit
sufficient time to locate a proxy to engage in the consent process. In some
states in which a delay in obtaining consent increases the risk of patient
harm, physicians may perform procedures without obtaining explicit consent. Implied consent assumes that most individuals would assent
to be treated in this situation; however, forgoing the usual dialogue that
is the underpinning of the consent process represents another potential encroachment
on patient autonomy.
Despite the difficulties presented by critical illness and the vulnerability
of patients in the ICU environment with regard to consent for treatment, there
is a paucity of medical literature concerning this topic. We recorded the
number of invasive procedures, along with the frequency with which they were
performed with informed consent, performed on all patients admitted to our
ICU during a 2-month period. We also analyzed the understanding by the individual
granting consent of the risks and benefits of these procedures. We repeated
these analyses during a subsequent 2-month period after changing the process
of informed consent via (1) adoption of a universal consent form for procedures,
(2) implementation of an educational initiative directed toward ICU clinicians,
and (3) implementation of formal proxy education through use of a handout.
We hypothesized that although consent for invasive procedures is not obtained
universally, a set of simple interventions would decrease the number of procedures
performed without informed consent, while preserving consenter understanding.
The study was conducted in a 16-bed medical ICU at a tertiary care teaching
hospital. This ICU is a closed unit, with all care directed by critical care
specialists. The physician team changes monthly and consists of 2 attending
physicians (each for 2 weeks) and 8 to 10 resident and fellowship trainees.
To survey opinions of the medical ICU team (nurses, house staff, and
attending intensivists), a baseline questionnaire covering 25 commonly performed
procedures was administered. For each procedure, team members were asked if
written consent was currently expected, as well as whether it should be mandatory
in the answering individual's opinion.
From November 1, 2001, until December 31, 2001, we identified all procedures
performed during the first 21 days of each patient's medical ICU stay. Procedures
recorded included placement of arterial catheter, central venous catheter
(CVC), pulmonary artery catheter (PAC), and peripherally inserted central
catheter; bronchoscopy; thoracentesis; paracentesis; lumbar puncture; gastrointestinal
tract endoscopy; blood transfusion; human immunodeficiency virus testing;
chest tube placement; and tracheostomy. Procedures were identified daily through
review of nursing and physician charting, query of physicians and nurses,
and direct examination of patients, seeking evidence of new procedures performed
(eg, a new catheter in place). The individual performing each procedure was
recorded. We also sought a written consent form in the medical record, as
well as the relationship to the patient of the person providing consent. Finally,
for all procedures without a written consent form, the reason for this was
solicited from the operator.
Individuals who gave consent were asked to complete a multiple-choice
questionnaire assessing comprehension of the consented procedure in 4 domains:
(1) the indication for the procedure; (2) a description of what the procedure
entailed; (3) the right to refuse the procedure; and (4) potential complications.
Adequate comprehension was prospectively defined as the consenter being able
to indicate (1) at least 1 reason the procedure was indicated; (2) a description
of what the procedure entailed; (3) knowledge of the right to refuse the procedure;
and (4) at least 3 potential complications associated with the procedure.
Demographic information obtained included age, sex, race, ICU diagnosis,
ICU length of stay, hospital length of stay, disposition after discharge from
the ICU, Acute Physiology and Chronic Health Evaluation II (APACHE II) score,11 use of mechanical ventilation, and mortality. This
study was approved by the institutional review board at the University of
Chicago, Chicago, Ill.
Following the initial 2-month baseline period of data collection, a
2-month intervention period ensued from March 1, 2002, to April 30, 2002.
During this time, we developed a 3-part intervention process, which included
a universal consent form that was approved by the institutional review board,
the forms committee, and the legal and ethics staff of the hospital (form
available from author on request). With this universal consent form, a patient
or proxy was able to give advanced written permission for 8 commonly performed
procedures (arterial catheter, CVC, PAC, or peripherally inserted central
catheter placement; lumbar puncture; thoracentesis; paracentesis; and intubation/mechanical
ventilation) in the ICU if necessary during the course of treatment. The second
part of the intervention was an attachment to the universal consent form describing
each procedure along with commonly associated complications. This information
was derived from current patient handouts available from the American Thoracic
Society, and sought to improve the education of patients and proxies regarding
these commonly performed ICU procedures.12 These
same informational handouts were available in the waiting area of the ICU.
If a patient or proxy refused to sign the universal consent form, the standard
practice of obtaining consent for each individual procedure was followed.
We prospectively defined improvement in the process of informed consent as
a higher incidence of documented informed consent with similar or improved
For the third part of the intervention, new residents and fellows were
introduced to the universal consent form as part of an orientation to the
ICU that occurred on their first day. The universal consent form was also
discussed with all nursing staff and was available for each new patient to
the ICU. This third part of the intervention—the clinician educational
component—consisted of a 15-minute discussion of the new consent form
and the rationale for its implementation. For the house staff, this was performed
by the ICU attending physician each month; for the nurses this was implemented
by one of the authors (A.P.), who is a clinical nurse educator. We subsequently
collected data on all admissions and procedures performed over another 2-month
period (March 1, 2002, until April 30, 2002). All data were collected daily
by 2 of the investigators (N.D., A.P.). These investigators were not privy
to the results of the data analyses until the study was complete.
Continuous demographic data were recorded as the mean (SD) or median
(interquartile range), as appropriate. Continuous demographic data from the
baseline and intervention periods were compared using the t test or the Mann-Whitney rank sum test, as appropriate. Categorical
data were compared between the baseline and intervention periods using χ2 analysis or the Fisher exact test. For those patients with multiple
ICU admissions, only the first ICU admission was analyzed. Based on our preliminary
observations before the study began, we noted that informed consent was obtained
for approximately 50% of procedures performed. With a goal of a 20% improvement
to 70%, with an α of .05, and a desired power of 90%, we estimated that
a sample size of 134 patients would be needed during each period. All statistical
analyses are 2-tailed and were performed using SPSS statistical software (Version
8.0, SPSS Inc, Chicago, Ill). A P value of less than
.05 was considered significant.
Seventy-three physicians and nurses were queried regarding opinions
of those procedures requiring written informed consent. Of those queried,
45 (62%) believed consent was required for arterial catheters; 71 (97%) for
CVCs; and 70 (96%) for PACs. Only 19 (26%) believed replacing an existing
arterial catheter (37 [51%] for a CVC) via a new procedure at a different
site required new written informed consent (Table 1).
One hundred twenty-five consecutive patients admitted to the medical
ICU were evaluated during the baseline period, while 145 consecutive patients
were evaluated during the intervention period. There were no significant differences
in any of the demographic variables collected except for a higher incidence
of mechanical ventilation in the baseline group (49.6% [62/125] vs 34.5% [50/145];
absolute difference = 15.1%; P = .02). Participant
characteristics are summarized in Table
During the baseline period, 53.1% (155/292) of procedures were performed
with written informed consent. Written consent was provided by 98 individuals,
including 27 patients who consented for themselves. The distribution of the
procedures and incidence of informed consent are shown in Table 3. Although blood transfusion (46/47; 98%), gastrointestinal
tract endoscopy (21/22; 95%), and bronchoscopy (11/11; 100%) all had high
percentages of written consent obtained, they only accounted for 80 (27.4%)
of 292 procedures performed. Of the 137 procedures performed without written
consent, 127 (93%) were for arterial catheters, CVCs, or PACs, despite the
baseline opinions expressed in our survey that it was expected and reasonable
for consent to be obtained (Table 1).
Of the procedures performed without written consent during the baseline
period, the most common explanations ICU physicians gave were emergency (n
= 34); consent not deemed necessary for a given procedure (n = 35); and no
proxy present (n = 11). Additionally, 20 procedures were performed without
a note in the medical record and we were unable to determine the reason for
forgoing written consent.
During the intervention period, with the availability of the universal
consent form, 90.5% of procedures (308/340) were performed with written consent.
This represented a significant difference compared with the baseline period
(53.1% vs 90.5%; absolute difference = 37.4%; P<.001).
Written consent was provided by 125 individuals, including 53 patients who
consented for themselves (P = .03 vs baseline period). Table 3 outlines the details for procedures
Of the 32 procedures performed without written consent during this time,
the most common explanations were emergency (n = 7) and no proxy present (n
= 5). Additionally, 16 procedures were performed without a note in the medical
record and we were unable to elicit the reason for waiving consent.
Of the 125 consenters, 83 received the universal consent form and 82
signed it. Of the 82 patients for whom universal consent was obtained, 75
actually underwent at least 1 procedure covered by this consent form.
Sixty (61.2%) of 98 consenters completed the procedure questionnaire
during the baseline period. The reasons the questionnaire was not completed
included death of patient (n = 6), participation refusal (n = 13), failure
to return the questionnaire after receiving it from the research team (n =
14), and inability of the research team to locate the proxy consenter (n =
5). Of the 60 persons completing the questionnaire, all demonstrated knowledge
of the indication for the procedure; the description of what the procedure
entailed; and the option to refuse. Forty-nine (81.7%) of 60 respondents were
able to recognize at least 3 potential complications related to the procedure.
Accordingly, we noted an 81.7% comprehension rate. Proxies consented for 71.6%
(111/155) of procedures.
Fifty-nine (71.0%) of the 83 individuals who received the universal
consent form also completed the procedure questionnaire. The reasons the questionnaire
was not completed in the other 24 included death of patient (n = 2); participation
refusal (n = 9); not returning questionnaire after receiving it from the research
team (n = 8); proxy not located (n = 3); and language barrier (n = 2). Of
the 59 persons completing the questionnaire, all demonstrated knowledge of
the indications for the procedure; the description of what the procedure entailed;
and the option to refuse the procedure. Fifty-three (89.8%) of 59 respondents
were able to remember at least 3 potential complications associated with the
procedure. Accordingly, we noted an 89.8% comprehension rate, similar to the
81.7% rate noted in the baseline period (absolute difference, 8.1%; P = .31). Proxies consented for 65.6% (202/308) of procedures.
In comparison with the baseline period, the absolute difference between baseline
and intervention is 6.0% (P = .23). During the baseline
period, 28% of consents for procedures were granted by patients compared with
34% during the intervention period. A spouse provided 20% of the consents
during the baseline period compared with 19% during the intervention period.
Other proxy consents were granted by children (22% during baseline vs 32%
during intervention), parents (8% vs 6%), siblings (18% vs 4%), individual
granted power of attorney (1% vs 1%), and other (2% vs 4%).
This investigation represents the first detailed description of the
process of informed consent for invasive procedures during the care of critically
ill patients. Proxies of ICU patients often experience anxiety and depression.
Precise, widely accepted guidelines for providing consent in this environment
do not exist, and there is diversity of opinion among clinicians as to which
procedures require consent. This diversity was reflected in our survey data.
Few clinicians thought that warming/cooling blanket use, peripheral intravenous
insertion, or bladder catheter placement required consent, yet all of these
procedures have indications, risks, and benefits. It is likely that most clinicians
view these procedures as a part of the general management occurring in the
ICU for which implicit consent exists. Our hospital policy states that informed
consent is required for "all invasive procedures performed outside the operating
room where there is more than minimal risk to the patient."13,14 Certainly
not all clinicians will agree on what constitutes "more than minimal risk."
Perhaps such vagaries partly explain the varying viewpoints and behaviors
Other procedures, such as blood transfusion, CVC insertion, thoracentesis,
lumbar puncture, and endoscopy, are viewed as almost invariably requiring
informed consent by our study clinicians. However, obtaining consent for these
procedures during our baseline study period was quite variable. This may relate
to a number of factors operative in the ICU environment. The first factor
is the frequent inability of the critically ill patient to participate in
the informed consent process due to the high incidence of delirium. In our
study, patients gave consent for only 34% of all procedures for which informed
consent was obtained. Although not statistically significant, a greater percentage
of procedures were consented to by patients during the intervention period,
which is consistent with a greater number of patients participating in the
consent process. Thus, procedures such as CVC insertion, which is often performed
on an emergent basis, may afford the clinicians a relatively short period
to contact and obtain consent from proxies. Without this consent, clinicians
in our study performed the procedure as a "medical emergency." Nevertheless,
our intervention substantially increased the incidence of consent obtained
for arterial catheters, CVCs, and PACs. This is particularly germane given
the high percentage of clinicians who deemed consent for such procedures to
be appropriate in our baseline questionnaire. It is also possible that a different
aura attends certain procedures that compel clinicians to make greater efforts
to obtain consent. This might be true of blood transfusions, given medical
and public concerns about risks therein related.
Regardless of the potential explanations for infrequent consent for
certain procedures, we thought that the observed rate of consent for certain
procedures was much lower than desired. We hypothesized that this might be
improved by instituting a standardized, comprehensive approach to the consent
process. We developed a universal consent form and procedure information handout,
seeking to improve patient and proxy comprehension without increasing ICU
physicians' burden. In considering the nature of the ICU environment, a prospective
approach to the potential need for invasive procedures seemed appropriate.
We reasoned that providing patients and proxies with information about procedures
at the time of admission to the ICU and obtaining consent for some of the
procedures that might be performed would facilitate their understanding of
these potential elements of care and would allow a greater number of proxies
to participate in important decisions regarding care.
Interestingly, the universal consent form was discussed with and signed
by only 57% of all proxies. Nevertheless, the majority of the actual procedures
performed during the intervention period were performed under the auspices
of this form. This suggests that physicians may be able to predict which patients
will require invasive procedures and thus are able to anticipate the need
for informed consent. Nevertheless, 36 procedures (20 baseline and 16 intervention)
occurred with neither consent nor documentation of the procedure. Perhaps
increased attention toward the importance of procedure documentation may reduce
this number further.
Given the limitations imposed by patient incapacity, the unpredictability
of critical illness, and the often complex mix of family members, friends,
and partners seeking to represent the patient's wishes, and assuming explicit
laws regarding substituted judgment even exist in the state in question, how
can informed consent in the ICU be used in the most effective manner? Our
intervention significantly increased the number of procedures for which written
consent was obtained, while preserving patient and proxy comprehension as
assessed by questionnaire. Our intervention was multifaceted in that it included
a universal consent form, proxy education, and clinician education. However,
it is not possible to know the individual impact of each intervention. Nevertheless,
this multifaceted approach was relatively simple, requiring a brief didactic
session with clinicians and written information for proxies. The universal
consent form, once constructed, was simple to use and well received by most
patients and/or proxies. Because it is conceivable that the universal consent
form could discourage some physicians from participating in regular communication
with the patient's proxy by removing the requirement for discussing each procedure
individually, data regarding patient and proxy satisfaction with the process
would have been interesting. It is notable, however, that only 1 of the 83
individuals presented with the universal consent form refused to sign it.
A potential limitation of anticipatory consent is that it does not allow for
alterations in the risk-to-benefit ratio of particular procedures as the patient's
condition changes. In these situations—for example, when a procedure
ordinarily contraindicated in the patient's present condition should, in the
treating physician's judgment, be performed anyway—one should ideally
obtain consent again.
Our study has several important limitations. Since it was conducted
at a single center, the practice of informed consent and the attitudes of
clinicians toward this process may be dissimilar to those of other institutions,
particularly in other countries in which the process of informed consent may
be different than in the United States. Our study design also did not allow
us to determine precisely the cause of the increased rate of informed consent
during the intervention period. Although the increase in rate of procedures
associated with the universal consent form suggests that this instrument was
responsible for the improvement, increased awareness through the educational
initiative or even a Hawthorne effect are also possible explanations. Similarly,
our study design may not be as rigorous as a randomized controlled trial.
However, the latter design could conceivably increase the Hawthorne effect
because the same clinicians would participate in 2 different consent processes
simultaneously. In our design, each period comprised at least 2 different
physician teams, allowing for a diverse array of clinicians participating
in the consent process. Lacking validated instruments for assessing patient
and proxy comprehension of procedures performed in the ICU, we cannot be certain
that our tools possessed the sensitivity to detect differences in understanding.
Further studies are needed to more reliably develop and validate tools for
assessing consenter comprehension. Finally, future studies that would include
assessment of patient or proxy satisfaction with the consent process are warranted.
We have shown that education of clinicians, patients, and proxies regarding
the process of informed consent can improve this process in critically ill
patients. In addition, by providing physicians and nurses with a standard
consent form and encouraging routine distribution of this form to patients
and/or proxies at admission to the medical ICU, we were able to significantly
increase the frequency with which informed consent was obtained for invasive
procedures. A patient or proxy's understanding of procedures was excellent
at baseline and was not compromised by routine distribution of this new instrument
for guiding informed consent.