Intensive Care Medicine

, Volume 35, Issue 10, pp 1703–1712

Patients’ preferences for enrolment into critical-care trials

  • Damon C. Scales
  • Orla M. Smith
  • Ruxandra Pinto
  • Kali A. Barrett
  • Jan O. Friedrich
  • Neil M. Lazar
  • Deborah J. Cook
  • Niall D. Ferguson

DOI: 10.1007/s00134-009-1552-y

Cite this article as:
Scales, D.C., Smith, O.M., Pinto, R. et al. Intensive Care Med (2009) 35: 1703. doi:10.1007/s00134-009-1552-y



Most critically ill patients are incapable of providing informed consent for research.


We sought to determine patients’ preferences for different consent frameworks for enrolling incapable patients into critical-care trials.


Prospective observational and structured interview study.


Five university-affiliated hospitals in Ontario.


Two-hundred and forty consecutive capable and consenting survivors of critical illness.


Participants considered four frameworks for enrolling incapable patients into clinical trials using a baseline scenario and three permutations for: risk (very low vs. high), treatment type (new vs. currently available), and availability of substitute decision-maker (yes vs. no).

Measurements and main results

For each scenario, patients chose their preferred framework and rated the acceptability of each framework using a seven-point Likert scale. Most (180/240; 76%) patients selected “consent by substitute prior to enrolment” as their preferred framework; this also received the highest baseline acceptability ratings (“acceptable” or “highly acceptable” 207/240; 87%). Modifying risk or treatment type did not substantially change these ratings. A minority of patients rated delayed consent as unacceptable or highly unacceptable in both the baseline scenario (48/240, 20% delayed to substitute; 57/240, 24% delayed to patient) and when a substitute was unavailable (34/240; 15%).


Most survivors of critical illness found the usual practice of obtaining informed consent from a substitute decision-maker prior to enrolment in a clinical trial to be acceptable. Nearly half of patients considered foregoing informed consent to be unacceptable, whereas a minority considered enrolment followed by delayed consent to be unacceptable even when a substitute was unavailable. These approaches should, therefore, only be considered when deviating from the usual practice of obtaining consent from a substitute decision-maker is truly justified, such as where treatments being tested need to be delivered as soon as possible in order to be effective.


Critical care Intensive care Informed consent Decision making Proxy Research design Ethics Research 


To reduce morbidity and mortality amongst critically ill patients, clinical research is imperative [26]. Because of severity of illness and/or sedation, however, most patients in the intensive-care unit (ICU) are incapable of providing first-person informed consent [12], which is the usual method for obtaining permission for enrolment in clinical studies [6, 24, 25]. This situation creates a tension between protecting the autonomy of these vulnerable patients, while at the same time ensuring that they are provided with the opportunity to participate in research that might help advance knowledge and improve care for similar patients in the future [31].

Most ethical analyses have concluded that the systematic exclusion of incapable patients from clinical research is undesirable because it violates the principle of beneficence, it makes it difficult or impossible to advance knowledge about the care of these patients, and could be construed as a form of discrimination against this vulnerable population [13, 26, 29, 43]. As a consequence, different frameworks have been proposed to enable research despite the inability to obtain first-party consent from incapable ICU patients [6, 7, 22, 23]. These include:

  1. 1

    consent by a substitute decision-maker (SDM) prior to enrolment;

  2. 2

    study enrolment followed by delayed consent by an SDM when one becomes available;

  3. 3

    study enrolment followed by delayed consent by the patient when they become capable; and

  4. 4

    study enrolment with no formal consent obtained.


In practice, the approaches used are determined by the nature of the study in question, and by local norms, guidelines, and common law [3, 18, 31, 33], and the interpretation of each of these by research ethics boards. However, patients’ preferences for these consent frameworks are unknown. We conducted a multicentre study to examine these preferences based on the assumption that such patients should not be systematically excluded from research.


A more detailed description of our methods is available in an online supplement.


Between August 2005 and August 2006 patients from five Ontario teaching hospitals who survived to ICU discharge were screened for study eligibility. Patients were considered potentially eligible if they demonstrated capacity to provide consent for the interview, but were excluded if they were not fluent in spoken English, were age ≤16 years, died or were discharged from hospital prior to demonstrating capacity, or were previously enrolled in this study.

Structured interview

We conducted structured interviews using a standardized script that was pre-tested and evaluated for ease of use, face and content validity, and comprehensibility by a group of nurses, physicians, and survivors of critical illness. The results of pre-testing and a copy of the standardized script are presented in the online supplement. We asked participants to consider a scenario where they were critically ill and incapable of providing consent for a randomized clinical trial. The participants were informed that the study under consideration in the scenario had been carefully reviewed and approved by a research ethics board, and was a randomized, placebo-controlled trial with a very low risk profile (<1/1,000 of serious complications); an SDM was available. We asked participants to first select which one framework for obtaining consent they would prefer for the scenario (Table 1), and then asked them to rate the acceptability of each consent framework using a 7-point Likert scale (1 = highly unacceptable to 7 = highly acceptable). We repeated this process for the following three permutations of the baseline scenario:

  1. 1

    increased risk (<1/20 instead of <1/1,000);

  2. 2

    treatment type (testing two currently available treatments instead of a new treatment); and

  3. 3

    availability of SDM (unavailable instead of available).

Table 1

Approaches for obtaining consent to enroll incapable patients in clinical research

Approaches evaluated in this study

  Consent from a substitute decision-maker prior to enrolment

  Enrolment followed by delayed consent by a substitute decision-maker

  Enrolment followed by delayed consent by the patient when they become capable

  No formal consent by patient or SDM required following review and approval by research ethics board

Approaches not evaluated in this study

  No research while patient is incapable

  Advanced or prospective consent obtained before the onset of critical illness


The primary outcome was the frequency distribution of participants’ preferred consent framework for the baseline research scenario. Secondary outcomes included:

  • the frequency distribution of the preferred consent framework for each permutation of the baseline scenario;

  • the proportion of participants finding each consent framework to be “acceptable” to “highly acceptable” (Likert scores = 6 or 7) and “unacceptable” to “highly unacceptable” (Likert scores = 1 or 2) in each scenario; and

  • the proportion of participants who changed their baseline preference with each permutation of the study scenario.


Using SAS version 9.1 (Cary, NC), we used the Student t test or Mann–Whitney U test, where appropriate, to test for differences in continuous variables between enrolled and non-enrolled patients. Bowker’s test of symmetry was used to compare proportions of patients selecting each consent framework across different study scenarios [8]. We also examined whether preference for “consent by SDM prior to enrolment” in the baseline scenario versus other consent frameworks was associated with participant characteristics using the chi-squared test or Fisher exact test for dichotomous variables and the Student t test for continuous variables. For all comparisons, a p value of <0.05 was considered to be statistically significant.


This study was approved by the Research Ethics Boards of all participating hospitals. Informed consent was obtained from all participants prior to the interview.


A total of 1,402 patients were discharged alive from participating ICUs during the study period, and 1,375 (98%) were screened for capacity a median of two times (IQR 1–3) after ICU discharge until deemed capable, hospital discharge, or death. Only seven patients (2%) were not screened because they were discharged or transferred prior to the first follow-up visit. Of 276 patients (20%) deemed capable, 240 (87%) agreed to be interviewed (Fig. 1).
Fig. 1

Study enrolment

On average, participants were 50 (SD 18) years of age, the median APACHE II score was 12 (IQR 8–17.5), and 42% were female. We present the characteristics of enrolled and non-enrolled ICU survivors in Table 2. In general, non-enrolled patients were older, more seriously ill, and more likely to be medical patients. Additional socio-demographic characteristics of enrolled patients are presented in Table 2.
Table 2

Characteristics of enrolled versus non-enrolled screened patients



N = 240


N = 1,162

p valuea

Age, mean (SD)

50 (18)

58 (19)


Female, N (%)

102 (42)

480 (41)


APACHE II score, median (IQR)

12 (8–17.5)

14 (9–20)


Mechanical ventilation, N (%)

145 (60)

637 (55)


Surgical, N (%)

120 (50)

408 (35)


Diagnostic category



Respiratory, N (%)

45 (19)

200 (18)


Cardiovascular (%)

29 (12)

119 (10)


Neurological (%)

71 (30)

373 (33)


Gastrointestinal (%)

48 (20)

151 (13)


Renal/genitourinary (%)

7 (2.9)

45 (4.0)


Trauma (%)

14 (5.9)

120 (11)


Other (%)

25 (10)

125 (11)


SDM approached during patient’s ICU stay for research participation, N (%)

11 (4.6)

27 (2.3)


Highest education attained, N (%)




Post-graduate degree

18 (8)


Undergraduate university degree

42 (19)


College degree

42 (19)


High school (completed)

76 (34)


High school (not completed)

33 (15)


Identified substitute decision-maker, N (%)





124 (52)



49 (20)



34 (14)



10 (4)


Power of attorney for personal care

4 (1.7)



10 (4.2)


Born in Canada (vs. elsewhere), N (%)

147 (62)



Ethnic originc





65 (27)



136 (57)


Other European

71 (30)


East Indian

12 (5.0)



9 (3.8)



80 (33)


aComparison is between enrolled versus non-enrolled screened patients

bNot applicable

cPercentages do not total 100% because multiple responses were allowed


The interviewers rated the comprehension of the majority (90%) of participants as “probably adequate” (38, 16%) or “definitely adequate” (178, 74%). When the eight (3.6%) patients with “probably inadequate” or “definitely inadequate” comprehension were excluded, our results did not change and are not reported (online supplement). Similarly, very few patients (1.8–4.3%) were unable to correctly identify the modifications to the baseline scenarios; sensitivity analyses excluding these patients did not change our results (online supplement).

Preferred framework for obtaining consent for research

Most patients (180; 76%) selected “consent by SDM prior to enrolment” as their preferred framework for the baseline scenario. A minority of patients selected “enrolment followed by delayed consent by SDM” (21; 8.9%) or “enrolment followed by delayed consent by patient” (21; 8.9%) as their preferred framework, and only 14 (5.9%) patients selected “no formal consent” from the patient or SDM. None of the patients’ clinical or socio-demographic characteristics we tested were associated with choosing “consent by SDM prior to enrolment” over other frameworks (online supplement).

Preferred consent framework for scenario permutations

Figure 2 shows the distribution of preferred frameworks across each hypothetical scenario, and Fig. 3 shows the change, if any, from the preferred framework selected at baseline. Increasing the risk of harm did not change patients’ preferred consent framework (Bowker’s test of symmetry, p = 0.12), although three-quarters (75%) considered this change to the scenario to be important. For this scenario, more than half (13, 62%) of those who had selected “enrolment followed by delayed consent by SDM” changed their preference to “consent from SDM prior to enrolment”.
Fig. 2

Preferred consent framework according to study scenario. This figure shows the patients’ preferred consent framework across different study scenarios. The coloured bars depict the proportion of patients selecting each consent framework. Numbers on graphs are percentages

Fig. 3

Change from baseline for preferred consent framework across different scenarios. This panel shows the change from baseline for preferred consent framework across different scenarios (squares). The categories along the x-axis represent the patients’ preferred consent framework for the baseline scenario. The coloured bars show the patients’ preferred consent framework after the baseline scenario has been modified to consider a higher risk of complications with the study therapy; a comparison of two currently available treatments; or for a situation in which no SDM is available. Numbers on graphs are percentages

When the scenario involved the study of two currently available treatments instead of a novel treatment, most patients still preferred the “consent from SDM prior to enrolment” framework (77%, Bowker’s test of symmetry versus baseline scenario p = 0.38). More than half of the patients (63%) considered this difference from the baseline scenario to be important. Patients who changed their preferred framework from the baseline scenario most frequently chose “consent from SDM prior to enrolment” (Fig. 3).

In the final scenario an SDM was not available. The majority of patients (83%) considered this difference to be important. When participants were asked to choose between the two remaining frameworks that did not involve asking an SDM, most (175; 77%) chose “enrolment followed by delayed consent by patient” over “no formal consent” as their preference. Most patients (81%) who had originally selected “consent from SDM prior to enrolment” for the baseline scenario switched their preference to “enrolment followed by delayed consent from patient” (Fig. 3) when the SDM framework was no longer available.

Acceptability ratings

We present patients’ acceptability ratings for each of the different consent frameworks across scenarios in Fig. 4. The consent framework receiving the highest acceptability ratings across all scenarios was “consent from SDM prior to enrolment”; this framework was rated acceptable to highly acceptable by more than 80% of patients whenever an SDM was available. Conversely, “no formal consent prior to enrolment” was rated as unacceptable to highly unacceptable for these same scenarios by more than half of patients. In contrast, only a minority of patients found the options of delayed consent to be unacceptable or highly unacceptable. When no SDM was available, “enrolment followed by delayed consent by patient” received the highest acceptability ratings, whereas “no formal consent” prior to enrolment was still rated unacceptable to highly unacceptable by nearly half of patients (46%).
Fig. 4

Acceptability ratings of different scenarios. This figure shows the patients’ acceptability ratings for each consent framework across different study scenarios. Acceptability ratings were based on Likert scales ranging from 1 = highly unacceptable through 7 = highly acceptable. Coloured bars represent proportion of patients rating each framework as unacceptable (red; Likert 1, 2), neutral (yellow; Likert 3, 4, 5), or acceptable (green; Likert 6, 7). Numbers on graphs are percentages


We systematically interviewed 240 ICU survivors with a variety of underlying diagnoses and found that most preferred their SDM to be involved in decisions about their participation in clinical research should they be incapable of making such decisions themselves. This preference did not change substantially when the study involved therapy associated with a higher risk of complications, or when the study compared two currently available treatments instead of a new intervention versus placebo. If an SDM was not available, most participants still wished to be involved at some point in the decision-making process.

Our results support the recommendations of critical-care guidelines, regulatory bodies, and critical-care societies for conducting ethical research on critically ill, incapable patients, namely that consent for enrolling incapable patients into clinical trials should be obtained from an SDM [3, 26, 31, 33]. This is an important finding, because although this process is usual practice in many jurisdictions, recent reports have highlighted the potentially tenuous legal foundation upon which it is based [21, 27, 28, 39]. Policy-makers, research ethics boards, researchers, and clinicians should be reassured that close to 90% of survivors of critical illness endorsed obtaining consent for research from an SDM as acceptable or highly acceptable. These findings are also consistent with studies demonstrating patients’ preferences to involve SDMs in medical decision-making [4, 40]. It is also reassuring to note that a preference for SDM involvement was robust across the various clinical and socio-demographic data that we tested, including age, APACHE II score, sex, education level, and having been born in Canada versus elsewhere.

Our results reinforce current standards for conducting research involving incapable patients, but they should be used to inform rather than dictate research policy. The choice of consent framework should be made considering the nature of the intervention under investigation and ethical principles, although stakeholder opinion should be sought to avoid research policy decision making that is devoid of input from the persons under investigation. The principles outlined in the Declaration of Helsinki, considering the duty of the researcher to protect the life, health, privacy, and dignity of research subjects and to ensure they are adequately informed about their rights to accept or decline to participate in research should continue to form the basis of research policy and methods of trial enrolment [1].

We did not interview SDMs to determine their preferences or to ask them to predict patients’ wishes. Others have examined the ability of SDMs to accurately predict patients’ preferences for medical and research decisions, and found high rates of discordance [9, 11, 32, 37, 38, 41, 42]. Collectively, these studies suggest that although patients may prefer their SDM be involved in research decision making, SDM predictions of the wishes of incapable patients regarding research participation do not always reflect patient self-reports. These studies were limited, like ours, by the use of hypothetical scenarios and the fact that patients’ preferences in interviews and in real-world situations may differ.

The care of patients needing immediate interventions in emergency settings would never be improved if all clinical trials investigating these situations were required to use SDM consent. When obtaining SDM consent is impractical, because of the emergent nature of the condition or treatment under study, alternatives are necessary to ensure that these vulnerable patients are not disadvantaged by a dearth of research evidence to guide their care. In these situations, delayed consent frameworks are being used with increasing frequency [7, 22, 23]. Several pre-hospital, emergency, and ICU studies have now been conducted using enrolment followed by delayed consent from SDMs [5, 10, 14], or by waiving the need for consent completely [34, 36]. The timely administration of some critical care interventions (such as fluid resuscitation [35], and antibiotics [19]), is now recognized as an essential aspect of their effectiveness, with ideal intervention windows of only a few hours. While they are not truly emergent, late administration of such interventions may render them less effective, and if they are markedly delayed, the ethics and feasibility of obligatory prior consent from an SDM is questionable [20]. We view the timeliness of initiating research on a continuum, reflecting the reality of practice and reflecting what is known about treatment delays in adult critical care medicine.

Our study did not explicitly assess preferences for different frameworks when the timing for enrolment was very short. Instead, we asked respondents to consider a situation where no SDM was available, reasoning that this would be the usual reality in emergency settings where finding an SDM could lead to enrolment delays that would render the study treatment ineffective. The fact that most patients who survived critical illness viewed delayed consent in a neutral or favourable fashion, even when an SDM was available, provides evidence for researchers and research ethics boards to consider when evaluating the suitability of delayed consent for urgent or emergent interventions. Importantly, if contacting an SDM in a timely fashion is not possible, three-quarters of patients in this study preferred that delayed first-person consent be obtained once capacity is regained. Unfortunately, such delayed first-person consent may be difficult to obtain for studies characterized by high mortality rates or prolonged incapacity [17]. Waiving consent entirely in these situations was considered to be acceptable by only one-quarter of our patients, suggesting other consent frameworks should be used if possible for routine critical-care research.

Modifying the consent process may have dramatic effects on study enrolment and interpretation of study results [45]. For example, the requirement for SDM consent created a biassed sample of patients in the observational study of the North American Stroke Registry, which was ultimately abandoned on these grounds [44]. In France, omitting the requirement for a priori SDM consent more than doubled the rate of recruitment into a randomized clinical trial of a sepsis treatment [2]. Patients in our study overwhelmingly opted for involvement of their SDM, suggesting that deviations from the usual practice of obtaining SDM consent must be truly justified. Our results imply that researchers should be wary of waiving SDM consent simply to hasten recruitment of study participants.

Our study has limitations. Our purpose was to evaluate patients’ preferences for different consent frameworks for conducting clinical research in the ICU. Thus, we did not give participants the option of choosing that “no research be conducted on incapable patients”, because this is not a framework for consent, and it had proved confusing to participants during pilot testing. The principle that incapable patients should not be systematically excluded from research that has the potential to inform and/or improve their care or that of those in similar future situations is consistent with most ethical analyses [6, 13, 30, 43]. We did not collect information that would help us to understand the patients’ reasons for preferring the SDM framework. Participants had all survived a critical illness, and therefore may reflect more positively upon their ICU experience than patients in general, potentially predisposing them to view critical-care research favourably. Participants were younger, less seriously ill, and more likely to be post-operative than non-enrolled patients; however, these are the characteristics of those consecutive patients who demonstrated the capacity to participate. The consistency of our findings favouring the SDM framework across all scenarios provides reassurance that randomizing scenarios would not have altered our conclusions. Finally, we restricted our focus to studying common and straightforward consent frameworks, rather than nuanced alterations of these such as SDM consent followed by patient consent once capacity is regained, consent deferred to community [16], or consent waived by prospective opt-out registry [15].


We found that the usual practice in North America and Europe of obtaining informed consent from an SDM to enroll ICU patients into clinical trials was acceptable to most patients who survived critical illness. Because few participants considered enrolment followed by delayed consent to be unacceptable, this approach might be expanded for ICU trials in which timely initiation of the study intervention could actually affect the effectiveness of the intervention.


We thank Mardi Baum, Katie L. Chapman, Greer Kirshenbaum B.Sc, Shelley Kraus B.Sc, Amanda Y. Li B.Sc, Kristina Lutz B.Sc, Jonathan S. Rosenberg B.Sc, Mohana Ratnapalan, Jasmine Soliven, and Ellen McDonald RN for assistance with data collection; Andrea Tkaczyk RN for preparing the research ethics application at St Joseph’s Healthcare; Helen Michalopoulous RN, and Michael Ward MD for input into initial discussions on study concept and design; and Sean P. Keenan, MD, M.Sc, Ellen McDonald RN, Donald A. Redelmeier, MD, and Roxanne Ward RN for critical revisions to earlier versions of this manuscript. This study was funded by grants from the Physicians’ Services Incorporated Foundation and the Canadian Intensive Care Foundation. None of the funding agencies had any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or the preparation, review, and approval of the manuscript. Dr Scales served as a consultant for the Ontario Ministry of Health and Long-term Care Critical Care Strategy. Dr Friedrich is supported by a Canadian Institutes of Health Research Clinician Scientist Award. Dr Cook is a Research Chair of the Canadian Institutes of Health Research. Dr Ferguson is supported by a RCT Mentoring Program Award from the Canadian Institutes of Health Research. We also thank all the patients who consented to be interviewed for this study.

Conflict of interest statement

None of the authors has any conflict of interest to declare.

Supplementary material

134_2009_1552_MOESM1_ESM.doc (462 kb)
(DOC 461 kb)

Copyright information

© Springer-Verlag 2009

Authors and Affiliations

  • Damon C. Scales
    • 1
    • 2
  • Orla M. Smith
    • 3
    • 4
  • Ruxandra Pinto
    • 2
  • Kali A. Barrett
    • 2
  • Jan O. Friedrich
    • 1
    • 4
  • Neil M. Lazar
    • 1
    • 5
  • Deborah J. Cook
    • 6
    • 7
  • Niall D. Ferguson
    • 1
    • 5
    • 8
  1. 1.Interdepartmental Division of Critical Care MedicineUniversity of TorontoTorontoCanada
  2. 2.Department of Critical Care MedicineSunnybrook Health Science CentreTorontoCanada
  3. 3.Faculty of NursingUniversity of TorontoTorontoCanada
  4. 4.Critical Care and Medicine Departments, Li Ka Shing Knowledge InstituteSt. Michael’s HospitalTorontoCanada
  5. 5.Division of Respirology, Department of MedicineUniversity Health NetworkTorontoCanada
  6. 6.Departments of Medicine and Clinical Epidemiology and BiostatisticsMcMaster UniversityHamiltonCanada
  7. 7.Department of Critical Care MedicineSt. Joseph’s HospitalHamiltonCanada
  8. 8.Toronto Western HospitalTorontoCanada

Personalised recommendations