Abstract
Rationale: Limited cross-sectional data exist to characterize the challenges of enrolling critically ill patients into research studies.
Objectives: We aimed to describe recruitment practices, document factors that impact recruitment, and identify factors that may enhance future research feasibility.
Methods: We conducted a prospective, observational study of all critically ill adults eligible to participate in research studies at 23 Canadian intensive care units. We characterized eligibility events into one of five consent outcomes, identified reasons why opportunities to recruit were missed or infeasible, and documented decision maker’s rationale for providing or declining consent.
Measurements and Main Results: Patients made decisions for themselves in 8.9% of encounters. In 452 eligibility events, consent was not required in 14 (3.1%), missed in 130 (28.8%), infeasible due to operational reasons in 129 (28.5%), obtained in 140 (31.0%), and declined in 39 (8.6%). More than half (57.3%) of all opportunities to recruit patients were missed or infeasible, largely because of research team workload, limited availability, narrow time windows for inclusion, difficulties in contacting families, nonexistent substitute decision makers (SDMs), physician refusals, and protocols prohibiting coenrollment. The rationale for providing consent differed between patients and SDMs. Greater research coordinator experience and site research volume and broader time windows for inclusion were significant predictors of fewer declined consents.
Conclusions: A large gap exists between eligibility and the frequency with which consent encounters occur in intensive care unit research. Recruitment is susceptible to design and procedural inefficiencies that hinder recruitment and to personnel availability, given the need to interact with SDMs. Current enrollment practices may underrepresent potential study populations.
Although numerous editorials have commented on the challenges of implementing acute care research, data supporting these assertions in the intensive care setting are limited.
(1) Approximately 10% of critically ill patients make decisions regarding their research participation, with the vast majority of decisions made by surrogates. (2) A large gap exists between eligibility and the frequency with which consent encounters occur with substitute decision makers (SDMs) in intensive care unit (ICU) research. More than half of opportunities to recruit eligible patients are missed or infeasible. (3) Enrollment into ICU research studies is susceptible to research personnel availability, given the need to interact with SDMs, and to design and procedural inefficiencies that hinder recruitment. (4) The rationale for providing consent for research participation differs between patients and their surrogates. (5) Greater research coordinator experience and site research volume and broader time windows for inclusion may be important predictors of fewer declined consents.
The need to investigate practices that improve outcomes of critically ill patients through high-quality and ethically sound research is widely recognized (1). However, the design and implementation of clinical research in the intensive care unit (ICU) pose challenges because of the need to recognize eligibility quickly, obtain consent for patients who lack decision-making capacity, initiate interventions in a time-sensitive manner, operationalize protocols under emergency conditions, and study conditions with high mortality.
The requirement for informed consent for randomization and the collection of personal health information follows from the ethical principle of autonomy and is embedded in ethical codes and research legislation (2, 3). Most critically ill patients are unable to comprehend or communicate and therefore cannot exercise autonomous decision-making (4). Consequently, informed consent for their research participation is typically obtained from surrogate decision makers (SDMs) or individuals involved in making decisions for a person who lacks decision-making capacity. The representativeness of SDM consent for research participation has been questioned amid evidence suggesting that some SDMs hold views or make decisions that differ from those of patients and may not know the patient’s values and preferences (5–7). A systematic review found that SDMs predict patient’s treatment preferences with only 65% accuracy (8). Moreover, when approached for participation, decision makers are often overwhelmed by medical information, the ICU environment, and the imminent threat of losing their loved one.
Difficulties in recruiting eligible patients into ICU studies may reduce opportunities for patients to participate in research; prolong study implementation; delay identification of effective, ineffective, and harmful treatments; and limit the generalizability of study findings (9, 10). The use of alternative consent models (e.g., waived or deferred consent) in randomized controlled trials (RCTs) may enhance recruitment and timely study completion with limited withdrawals (11–13). Despite favorable attitudes toward their use in questionnaires administered to SDMs and the general public (14, 15), institutional review boards (IRBs) rarely approve them and individuals other than SDMs seldom provide consent for research participation (16). Although editorials have commented on the challenges of implementing acute care research (17), data supporting these assertions are limited. We conducted a multicenter, cross-sectional study to characterize research recruitment practices in Canadian adult ICUs, document factors that facilitate and impede recruitment, and identify potentially modifiable factors that may enhance future research feasibility. Some of the results of these studies have been previously reported in the form of abstracts (18, 19).
We conducted a prospective observational study of research recruitment processes in 23 community and tertiary care adult intensive care units (ICUs) across Canada between February 2009 and June 2009. Data collection occurred over a 1-month period in each ICU. We included critically ill adults who were eligible to participate in any critical care research study. We excluded patients who died or were transferred out before screening. Research ethics approval was obtained at all participating sites and consent for participation was not required.
Our primary objective was to characterize eligibility events into one of five categories, wherein (1) consent was not required (e.g., waived consent), (2) the opportunity for consent was missed (e.g., weekend, time ran out such that enrollment could not be completed in time), (3) research teams were unable to obtain consent for operational reasons (e.g., no SDM or next of kin, language barrier, coenrollment prohibited), or consent was (4) obtained or (5) declined; and to identify predictors of an SDM’s decision to decline consent.
Our secondary objectives were to describe the proportion of SDMs and patients approached for consent and the time interval between recognition of eligibility, being approached for consent, and a consent decision being made; and to identify reasons why eligible patients are missed, investigative teams are unable to obtain consent for operational reasons, and the rationale for patients and SDMs providing or declining consent. We compared consent outcomes based on (1) the identity of the person initially approaching decision makers (research coordinator vs. other), (2) the type of SDM (spouse vs. other), (3) investigator- versus industry-initiated RCTs, (4) RCTs versus other study designs, (5) prevention versus treatment studies (retrospectively collected), (6) pilot versus full-scale RCTs, (7) mechanical ventilation versus other RCTs, and (8) RCTs with narrow (≤24 and ≤48 h) and broader (>24 and >48 h) time windows for inclusion.
A one-page (consent not required, opportunity missed, or unable to consent for operational reasons) or two-page (consent obtained or declined) event was completed for each eligibility event. Site forms and study forms captured site-related and study-related information.
Descriptive statistics including mean and standard deviation (alternatively, median and interquartile range) and proportions were used to summarize continuous and binary outcomes, respectively. We compared continuous measures by Student t test (or Wilcoxon rank sum test for skewed distributions) and binary outcomes by chi-square test (alternatively, Fisher exact test for expected values < 5). The individual event was considered the unit of analysis. We used SDM postal code and the forward sortation function in SAS 9.2 (SAS Institute, Cary, NC) to determine income quintiles with lower values representing lower socioeconomic status (20). We identified the predominant reason why decision makers provided or declined consent.
We anticipated collecting data on 25 events per ICU and planned to conduct hierarchical linear modeling (21) to identify predictors of SDM decisions to provide or decline consent, using 18 prespecified variables. We recognized that this likely represented an overestimate given variability in ICU size, research capacity, and patient turnover. With a smaller number of consent outcomes, we prioritized 11 variables, through discussion, and used logistic regression with generalized estimating equations analysis to identify predictors (22). We report predictors of declined consent using odds ratios (ORs) and 95% confidence intervals (CIs) and considered P values less than 0.05 to be statistically significant.
There were 1,812 admissions in the 23 participating ICUs, of which 453 admissions (25.0%) occurred on weekends. We present the characteristics of eligible patients in Table 1. Most eligibility events occurred in medical-surgical (n = 278), medical (n = 85), or multidisciplinary (n = 78) ICUs. ICUs had an average of 17.4 ± 7.6 and 17.2 ± 7.5 beds open on weekdays and weekends, respectively, and 1.6 ± 0.6 full-time equivalent research coordinators. Research coordinators were available after hours for all studies on weeknights and weekends in 13 of 23 (56.5%) and 8 of 23 (34.8%) ICUs and, for specific studies, in 6 of 23 (26.1%) and 10 of 23 (43.5%) ICUs, respectively. General screening logs were maintained by 8 of 23 ICUs (34.8%). Although most ICUs had access to interpreters (69.6%) for consent discussions, only one provided research documents in a language other than English or French. Sites had an average of 19.0 ± 8.9 IRB members and 0.3 ± 0.6 seats designated for critical care. Most IRBs had representation from adult critical care (14/23 [60.9%]) or emergency medicine (8/23 [34.8%]) and permitted telephone (17/23 [73.9%]) or facsimile consent (17/23 [73.9%]). Research personnel from most ICUs (19/23 [82.6%]) were permitted to coenroll patients into more than one study. We excluded data from one ICU with nonsystematic screening and seven events (four events in which recognition of eligibility did not occur in the month of prospective data collection and three events in which patients were no longer eligible at the time of recognition).
| Demographic Data | Consent Not Required | Consent Missed | Unable to Consent | Consent Obtained | Consent Declined | Total |
|---|---|---|---|---|---|---|
| n (%) | 14 (3.1) | 130 (28.9) | 129 (28.5) | 140 (30.9) | 39 (8.6) | 452 |
| Mean age ± SD | 67.2 ± 19.5 | 66.6 ± 16.2 | 65.2 ± 15.8 | 64.0 ± 15.7 | 68.5 ± 12.9 | 65.6 ± 15.8 |
| Male, n (%) | 5 (1.1) | 86 (19.0) | 86 (19.0) | 75 (16.6) | 17 (3.8) | 269 (59.5) |
| Readmission, yes (%) | — | 7 (1.6) | 9 (2.0) | 10 (2.2) | 2 (0.4) | 28 (6.2) |
| Type of ICU, n (%) | ||||||
| Medical | — | 36 (8.0) | 26 (5.8) | 16 (3.5) | 7 (1.5) | 85 (18.8) |
| Surgical | — | — | — | 3 (0.7) | — | 3 (0.7) |
| Medical/surgical | 12 (2.7) | 71 (15.7) | 87 (19.2) | 85 (18.8) | 23 (5.1) | 278 (61.5) |
| Multidisciplinary | — | 23 (5.1) | 16 (3.5) | 31 (6.8) | 8 (1.8) | 78 (17.3) |
| Neurotrauma | 2 (0.4) | — | — | 5 (1.1) | 1 (0.2) | 8 (1.8) |
We recorded data on 452 eligibility events for 291 patients. In 14 of 452 (3.1%) events consent was not required, in 130 of 452 (28.8%) events consent was missed, and in 129 of 452 (28.5%) investigative teams were unable to obtain consent for operational reasons. Consent was declined in 39 of 452 (8.6%) and obtained in 140 of 452 (31.0%) of eligibility events. In three encounters patients were enrolled and consent was deferred to either the SDM or the patient. ICUs recorded a median of 11 (5, 23; range, 1–85) eligibility events over the study period.
Patients for whom consent was provided and declined were similar in mean age (64.0 ± 15.7 vs. 68.5 ± 12.9 yr; P = 0.11) and had similar organ dysfunction (Sequential Organ Failure Assessment [SOFA]) scores (6.5 ± 3.3 vs. 7.0 ± 3.6; P = 0.51). Consent was provided more often than it was declined by patients and surrogate decision makers from low (25/28 [89.3%] vs. 3/28 [10.7%]), middle (66/80 [82.5%] vs. 14/80 [17.5%]), and high (20/30 [66.7%] vs. 10/30 [33.3%]) socioeconomic groups with nonsignificant differences (P = 0.07) among groups. Research coordinators made initial contact alone in encounters in which consent was obtained (124/140 [88.6%]) or declined (31/39 [79.5%]) and physicians were infrequently involved in introducing research studies to decision makers. All consents were obtained in writing.
Consent was obtained or declined (n = 179) for 30 studies involving 147 patients (Table 2). Patients directly provided or declined consent in only 16 of 179 (8.9%) encounters. Conversely, spouses or partners functioned as SDMs in nearly half of all consent encounters (83/179 [46.4%]). For 95 and 22 patients, a single encounter occurred for which consent was obtained or declined, respectively. In 19 patients consent was obtained for 2 studies, in 5 patients consent was declined for 2 studies, and in 4 patients consent was both obtained and declined. For one patient each, consent was obtained and declined on three occasions. Decision makers were most often approached for a single study (133/179 [74.3%]; 106 consents obtained and 27 declined), and less often for participation in two studies (40/179 [22.3%]; 31 consents obtained and 9 declined) or three studies (6/179 [3.4%]; 3 consents obtained, 3 declined).
| Study Features | Obtained Consents (n = 140) | Declined Consents (n = 39) |
|---|---|---|
| Study design, n (%) | ||
| Full RCT | 112 (80.0) | 35 (89.7) |
| Pilot RCT | 10 (7.1) | 2 (5.1) |
| Observational study (no intervention) | 1 (0.7) | — |
| Observational study (with intervention) | 8 (5.7) | 2 (5.1) |
| Biological specimen collection | 7 (5.0) | — |
| Other | 2 (1.4) | — |
| Sponsorship, n (%) | ||
| Investigator-initiated | 122 (87.1) | 34 (87.2) |
| Industry-sponsored | 11 (7.9) | 3 (7.7) |
| Investigator–industry partnership | 7 (5.0) | 2 (5.1) |
| Time window for inclusion, n (%) | ||
| ≤24 h | 35 (25.0) | 6 (15.4) |
| ≤48 h | 40 (28.6) | 7 (17.9) |
| Not applicable | 35 (25.0) | 13 (33.3) |
| Active intervention, n (%) | ||
| Drug | 69 (49.3) | 19 (48.7) |
| Ventilation/weaning | 8 (5.7) | 2 (5.1) |
| Sedation | 17 (12.1) | 9 (23.1) |
| Nutrition supplements | 18 (12.9) | 5 (12.8) |
| Device | 5 (3.6) | — |
| Interview | 4 (2.9) | 2 (5.1) |
| Not applicable* | 4 (3.0) | — |
| Other | 15 (10.7) | 2 (5.1) |
| Control intervention, n (%) | ||
| Placebo | 5 (3.8) | 3 (7.7) |
| No intervention/comparator | 2 (1.5) | — |
| Another intervention/algorithm or protocol | 111 (79.3) | 33 (84.6) |
| Not applicable* | 22 (15.7) | 3 (7.7) |
| Approved consent models, n (%) | ||
| Patient or SDM | 126 (90.0) | 37 (94.9) |
| Deferred | 5 (3.6) | 1 (2.6) |
| Advance (patient or SDM) | 9 (6.4) | 2 (2.6) |
Compared with the least experienced coordinators, consent was declined significantly less often to experienced research coordinators in the second (OR, 0.04; 95% CI, 0.00, 0.50; P = 0.01) and fourth (OR, 0.05; 95% CI, 0.56, 0.77; P = 0.01) quartiles of experience, in centers performing more research (implementing ≥3 [median] vs. ≤2 studies) (OR, 0.05; 95% CI, 0.01, 0.52; P = 0.01), and for studies with > 24-hour time windows for inclusion (Table 3). Sensitivity analyses attributing not applicable time windows as within and outside of the 24-hour time window and excluding this variable supported our findings that research coordinator experience and research volume were significant predictors.
| Variable | Odds Ratio | Confidence Interval | P Value |
|---|---|---|---|
| Total SOFA (tertile: 2 vs. 1) | 0.94 | 0.07–12.74 | 0.96 |
| Total SOFA (tertile: 3 vs. 1) | 0.38 | 0.01–12.90 | 0.59 |
| Age (quartile: 2 vs. 1) | 0.47 | 0.02–9.34 | 0.62 |
| Age (quartile: 3 vs. 1) | 0.61 | 0.08–4.51 | 0.63 |
| Age (quartile: 4 vs. 1) | 0.46 | 0.05–4.45 | 0.50 |
| SDM preferred language (other vs. English) | 11.33 | 0.99–129.90 | 0.05 |
| Income (quintile: 2–4 vs. 1) | 1.25 | 0.03–55.47 | 0.91 |
| Income (quintile: 5 vs. 1) | 7.32 | 0.05–1,022.8 | 0.43 |
| Time window: (>24 vs. ≤24 h) | 0.14 | 0.03–0.57 | 0.006 |
| Years experience of FTE RCs (quartile: 2 vs. 1) | 0.04 | 0.00–0.50 | 0.01 |
| Years experience of FTE RCs (quartile: 3 vs. 1) | 0.12 | 0.00–3.56 | 0.22 |
| Years experience of FTE RCs (quartile: 4 vs. 1) | 0.05 | 0.01–0.56 | 0.01 |
| Person making initial contact: (RC vs. other) | 1.11 | 0.05–24.91 | 0.95 |
| Number of FTE RCs (tertile:2 vs. 1) | 5.04 | 0.13–190.43 | 0.38 |
| Number of FTE RCs (tertile: 3 vs. 1) | 5.30 | 0.08–339.64 | 0.43 |
| Number of studies (≥3 vs. ≤2 studies) | 0.05 | 0.01–0.52 | 0.01 |
| Design type (RCT vs. other) | 0.23 | 0.03–2.00 | 0.18 |
Only 15 patients, involved in 16 of 179 encounters (8.9%), functioned as their own decision makers. In encounters, patients provided and declined consent (14/140 [10.0%] vs. 2/39 [5.1%]; P = 0.53) with similar frequency. There were no significant differences in consent rates (obtained vs. declined) provided by spouses/partners (65/126 [78.3%] vs. 18/37 [48.6%]; P = 0.85), siblings (10/126 [7.9%] and 4/37 [10.8%]; P = 0.52), parents (12/126 [9.5%] vs. 0/37 [0%]; P = 0.07), and children (35/126 [27.8%] vs. 15/37 [40.5%]; P = 0.16) decision makers.
After recognition of eligibility, contact with decision makers was made after a median of 1.8 (0.75, 5.0) hours (mean, 8.8 h) and after 3.5 (1.0, 6.5) hours (mean, 12.8 h) for events in which consent was obtained and declined, respectively (P = 0.21). We did not identify differences in the median time from decision-maker contact to a consent decision being rendered (obtained, 0.5 [0.25, 1.50] vs. declined, 0.5 [0.25, 5.5]; P = 0.63) (mean, 2.9 vs. 5.7 h) or from eligibility recognition to a consent decision being rendered (obtained, 3.3 [1.5, 8.5] vs. declined, 4.5 [1.3, 25.8]; P = 0.16) (mean, 11.6 h obtained vs. 18.5 h declined).
More than one-half (57.3%) of opportunities to recruit critically ill patients eligible for research participation were either missed or not realized. Missed opportunities (n = 130) were primarily attributed to research team workload (39 [30.0%]), lack of available research staff (36 [27.7%]), expired time windows (27 [20.8%]), and decisions made not to approach decision makers (16 [12.3%]). In less than 10% of events, eligible patients were missed because of an oversight (3.8%), death (3.1%), or clinical (ICU- or patient-related) factor that impeded RC workload (2.3%).
Research staff were unable to obtain consent (n = 129) because of difficulties in contacting families (31 [24.0%]), because no decision maker existed (27 [20.9%]), physicians refused patient participation (23 [17.8%]), and trials prohibited coenrollment (19 [14.7%]). Less frequently, investigative teams experienced difficulty in establishing who the decision maker was (4.7%), SDMs were not able to attend the ICU (9.3%), IRBs prohibited coenrollment (5.4%), or a language barrier existed (3.1%).
Reasons for providing consent were obtained from patients (n = 9), SDMs (n = 107), and research coordinators (n = 24). Critically ill patients expressed a strong desire to help others (future patients, the medical community; 5/14 [35.7%]) and advance knowledge/science (6/14 [42.9%]). In contrast, SDMs directly or indirectly believed that their loved one could benefit from participation (81/126 [64.3%]) and less frequently expressed a desire to help others (22/126 [17.5%]) or to advance knowledge/science (14/126 [11.1%]). Few patients (1/14 [7.1%]) or SDMs (5/126 [4.0%]) held strong positive convictions regarding patient participation.
Only two patients involved in two events directly declined consent for research participation; one citing a fear of participation and the other, unable to deal with the request for participation at the time. Commonly cited reasons for 37 declined SDM consents (31 SDM opinions, 6 research coordinator opinions) were a desire not to change current treatment (8/37 [21.6%]), apprehension regarding participation (6/37 [16.2%]), uncertainty regarding the patient’s wishes (5/37 [13.5%]), an intuition that the patient would not want to participate (5/37 [13.5%]), and being too upset to contemplate participation (4/37 [10.8%]).
We found no difference in consent rates based on the person who was approached for consent and study design features (Table 4).
| Study Design Features | Obtained Consents [n (%)] | Declined Consents [n (%)] | P Value |
|---|---|---|---|
| Who approached | |||
| RC | 130 (79.3) | 34 (20.7) | 0.32 |
| Other | 10 (66.7) | 5 (33.3) | |
| Study type | |||
| Observational/other | 24 (88.9) | 3 (11.1) | 0.30 |
| Prevention | 57 (78.1) | 16 (21.9) | |
| Treatment | 59 (74.7) | 20 (25.3) | |
| Type of SDM | |||
| Spousal | 65 (78.3) | 18 (21.7) | 0.75 |
| Other | 61 (76.3) | 19 (23.8) | |
| RCT study sponsorship | |||
| Investigator initiated* | 112 (76.7) | 34 (23.3) | 0.99 |
| Industry sponsored | 10 (78.9) | 3 (23.1) | |
| Design | |||
| RCT | 122 (76.7) | 37 (23.3) | 0.18 |
| Other | 18 (90.0) | 2 (10.0) | |
| Type of RCT | |||
| Full | 112 (76.2) | 35 (23.8) | 0.57 |
| Pilot | 10 (83.3) | 2 (16.7) | |
| Nature of intervention in RCTs | |||
| Mechanical ventilation | 8 (80.0) | 2 (20.0) | 1.00 |
| Other | 114 (76.5) | 35 (23.5) | |
| RCTs with narrow time window | |||
| ≤24 h | 33 (84.6) | 6 (15.4) | 0.28 |
| >24 h | 60 (77.9) | 17 (22.1) | |
| Not applicable | 33 (70.2) | 14 (29.8) | |
| RCTs with broader time window | |||
| ≤48 h | 38 (84.4) | 7 (15.6) | 0.27 |
| >48 h | 55 (77.5) | 16 (22.5) | |
| Not applicable | 33 (70.2) | 14 (29.8) | |
In a multicenter cross-sectional study, we found that 57% of opportunities to recruit eligible patients are either missed or not realized because of operational reasons. From these, we identified several potentially modifiable barriers to patient recruitment including research team workload and availability, narrow time windows for inclusion, protocols prohibiting coenrollment, and physician refusals. More than 90% of consent encounters involved SDMs. We noted that the rationale for providing consent differed between patients and SDMs, with patients expressing more altruism and SDMs believing that their loved one could benefit from participation. We identified greater research coordinator experience and site research volume and broader time windows for inclusion as significant predictors of fewer declined consents. Few studies have quantified ICU recruitment challenges. In a prospective study evaluating associations among head of bed elevation, oral care, and ventilator-associated pneumonia in a medical–respiratory ICU, Grap and Munro found that only 42 (7.1%) patients were enrolled under SDM consent, with more patients being eligible but not enrolled (94 [15.8%]) or ineligible (457/593 [77.1%]) (23). In a retrospective study of a low versus traditional tidal volume trial, Glassberg and colleagues noted that patients with acute lung injury were less likely to be enrolled at public compared with university hospitals because of unavailable or absent SDMs (24). In a 1-month prospective study involving 2,698 admissions from 20 adult and 6 pediatric ICUs, Marshall and Cook found that 136 patients, representing 5% of ICU admissions, were recruited into studies (25). We found that whereas 31.0% (7.7% of ICU admissions) of all eligible patients were enrolled into studies, more than half (57%) were eligible but not enrolled. Disparities in estimates of eligible but not enrolled patients between studies may reflect differences in the definitions used, case mix, study design features, consent processes, or interventions. In addition, recruitment rates may be reported as a proportion of all eligible patients or ICU admissions, with the former being a more useful metric of recruitment efficiency.
We identified several potentially modifiable barriers to ICU recruitment. Featured prominently among the missed opportunities for consent were the workload and availability of research personnel. We also identified several study design and implementation considerations that may impede ICU recruitment including narrow time windows for inclusion, protocols that prohibit coenrollment, and physician refusals. Similar to our findings, the Randomized Evaluation of Normal versus Augmented Level (RENAL) Study found that lack of available research staff (7.3%) and physician refusals (5.2%) represented recruitment barriers (26). The Acute Renal Failure Trial Network (ATN) Study also noted the adverse impact of delayed participant identification (4.4%), physician refusals (2.7%), and competing trials (2.3%) on trial recruitment (27). In the ATN study, a modification to the time window for enrollment reduced patient losses from 6.6 to 2.3%. A randomized trial of prone ventilation versus supine ventilation was prematurely terminated because of difficulty in evaluating and enrolling patients with adult respiratory distress syndrome within a 48-hour time window and amid competing studies funded by pharmaceutical companies (28). Conversely, a prospective study of patients with traumatic brain injury highlighted that whereas the ability of research staff to contact families increased over time, surrogate consent rates did not (29). Although telephone consent may shorten the time from initial contact to a consent decision being rendered, its impact on consent rates remains to be determined (30). A trinational survey of members of two large ICU research consortia suggests that researchers endorse coenrollment into more than one RCT (31), with 52% of respondents acknowledging prior experience with coenrollment. In the Prophylaxis for Thromboembolism in Critical Care Trial, 713 of 3,746 (19.0%) were coenrolled in at least one other study (53.6% in a randomized trial, 37.0% in an observational study, and 9.4% in both) (32). Coenrollment occurred in our study in 50 events involving 20/119 (16.8%) patients and 11 studies. In 24/119 patients for which consent was obtained and an additional opportunity existed, coenrollment did not occur predominantly because trials and IRBs prohibited coenrollment, suggesting that opportunities to enhance recruitment through coenrollment may exist. Our findings also suggest that consent procurement may be a learned skill that improves with experience and volume. Despite evidence supporting families’ desire for research information, few ICUs in our study used posters or pamphlets to increase research awareness (33). The low visibility of ICU research may compound ICU recruitment challenges.
Unlike most settings, ICU research personnel are typically unable to interact directly with patients because of illness severity; the presence of delirium and comorbidities; and use of sedatives, analgesics, and life-sustaining treatments (34–36). Only 9% of encounters in this study involved patients providing first-party consent. Absent decision makers (15–40%) and refusals (18–27%) have been identified as factors that impeded recruitment into observational studies (23, 37). Two large renal replacement RCTs involving ICU patients identified consent procurement (11.2–21.4%) as the principal barrier to recruitment (26, 27). Notwithstanding, Scales and coworkers found that compared with alternative consent models (waived or deferred consent), ICU survivors favored SDM involvement in decisions regarding their research participation across scenarios (38). However, although SDMs were willing to be involved, they were often uncomfortable making decisions regarding research participation and desired input from others, especially physicians (39). Others opine that alternative consent models are needed and contend that although research processes involving vulnerable patients must be held to the highest ethical standards, they should not preclude patient access to potentially beneficial therapies when no effective therapy for life-threatening conditions exist (40). As a solution to complete reliance on SDM consent, investigators have proposed using alternative consent models for low-risk studies, hybrid consent models (prioritizing SDM involvement when available), and preemptive consent (9, 16, 41). Although RCTs using these models have demonstrated enhanced recruitment and limited withdrawals, no study has evaluated their acceptability to SDMs (12, 13, 42). A large gap exists between eligibility and the frequency with which consent encounters occur with SDMs in ICU research. This gap infrequently exists outside the acute care setting, where eligibility and encounters are more closely approximated and are likely compounded by research personnel availability and workload.
Although discordance between patient and SDM decisions regarding research participation has been documented (6, 8, 43), reasons for these discrepancies are not well delineated. We noted that whereas patient motives for participation were altruistic, espousing a desire to help others and advance science, SDM motives focused on a belief that their loved one could benefit from participation. Common reasons why SDMs decline consent in this study included a desire not to change current treatment, apprehension about participation, uncertainty regarding the patient’s wishes, and a belief that the patient would not want to participate. In a questionnaire administered to 123 SDMs in a single center, of whom 76% (68 consent providers and 27 consent decliners) responded, Mehta and coworkers found that common reasons for providing consent included the potential for research to help others (91%), recognition that research was important for medical progress (88%), and trust in the medical team (87%). SDMs declined consent for participation because they were too anxious to consider research (67%), feared patients would receive experimental treatment (37%), and were concerned about study risks (33%) (44). Discordant findings between these studies may reflect different response formats used (check one vs. all that apply, respectively) and options provided. Patient and surrogates consider and weigh multiple factors in rendering consent decisions (45). Additional research is needed to clarify SDM perceptions of the link between research participation and patient outcomes (46).
Several potentially modifiable barriers to ICU recruitment exist. Increased coordinator workload and lack of available staff highlight the need for infrastructure funding to maximize limited opportunities to discuss research participation for eligible patients. Moreover, careful consideration should be given by investigators and research ethics boards to study design (e.g., narrow time window, protocols prohibiting coenrollment) and implementation (physician refusal) issues that may hinder recruitment. The impact of novel approaches including cross-site and within-site research personnel coverage; use of telephone, facsimile, and electronic consent, preemptive consent, alternative consent models, or hybrid consent models for low-risk studies and approaches to introduce research to SDMs on recruitment should be formally evaluated. Our multivariate analysis suggests that successful consent procurement by research personnel may be a learned skill that is favorably impacted by experience and workload.
Strengths of this study include the prospective nature of the data collection; the multidisciplinary involvement in data form development; and collection of patient, SDM, consent, study, and institutional data. The national and cross-sectional design differs from previous studies evaluating recruitment in single-center, cohort studies and within individual RCTs. The scope of our study, involving 29 studies with eligibility events and 23 adult ICUs, enhances the generalizability of our findings. Our study also has limitations. Although we endeavored to prospectively identify all eligibility events from the time of ICU admission, given the minimum data collected for patients who were missed and for whom we were unable to obtain consent, we cannot distinguish between patients who were prospectively versus retrospectively deemed eligible for study participation. The time of eligibility recognition was at the discretion of research coordinators. Second, time intervals among eligibility recognition and contact and consent varied due to several factors including preconsent, consent models, and staged approach for inclusion. Third, the number of encounters for which consent was obtained or declined in this study was moderate, necessitating prioritization of variables for inclusion in the multivariate analysis. Fourth, although we identified differences in the rationale for consent decisions, some assessments were provided by research personnel. Fifth, our findings reflect the studies in operation in participating centers during the data collection period, with two trials contributing a large number of eligibility events, and may not be generalizable to other settings. Similarly, identified predictors may reflect our prioritization and categorization of variables and the studies included in the regression analyses. Finally, we were unable to ascertain or classify 41 postal codes for SDMs, and did not include one prespecified variable (redundant) in the regression analysis; 65 events remained.
More than one-half of opportunities to recruit eligible patients were either missed or not realized because of several potentially modifiable factors including research team workload and availability, narrow time windows for inclusion, protocols that prohibit coenrollment, and physician refusals. Given the need to interact with SDMs for consent, ICU research may be particularly susceptible to personnel availability and design and procedural inefficiencies that hinder recruitment. Consequently, current enrollment practices may underrepresent study populations. Research on the process of conducting ICU research is required to evaluate strategies to enhance participation.
The authors thank the research coordinators of the Canadian Critical Care Trials Group who collected data for this study, and the patients and surrogate decision makers who participated in the Consent Study. The authors also thank Ms. Leena Rizvi for assistance in preparing the data for analysis.
| 1. | Cook D, Brower R, Cooper J, Brochard L, Vincent JL. Multicenter clinical research in adult critical care. Crit Care Med 2002;30:1636–1643. |
| 2. | National Institutes of Health, Office of Human Subjects Research. Regulations and ethical guidelines. Directives for human experimentation: the Nuremberg code [accessed 2012 Mar 26]. Available from: http://ohsr.od.nih.gov/guidelines/nuremberg.html |
| 3. | National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research. Belmont Report: ethical principles and guidelines for the protection of human subjects of research 3. 44 Federal Register 23192 (1979). |
| 4. | Luce JM. Is the concept of informed consent applicable to clinical research involving critically ill patients? Crit Care Med 2003;31(3, Suppl):S153–S160. |
| 5. | Upadya A, Muralidharan V, Thorevska N, Amoateng-Adjepong Y, Manthous CA. Patient, physician, and family member understanding of living wills. Am J Respir Crit Care Med 2002;166:1430–1435. |
| 6. | Coppolino M, Ackerson L. Do surrogate decision makers provide accurate consent for intensive care research? Chest 2001;119:603–612. |
| 7. | Bolmsjö I, Hermerén G. Interviews with patients, family, and caregivers in amyotrophic lateral sclerosis: comparing needs. J Palliat Care 2001;17:236–240. |
| 8. | Shalowitz DI, Garrett-Mayer E, Wendler D. The accuracy of surrogate decision makers: a systematic review. Arch Intern Med 2006;166:493–497. |
| 9. | Burns KE, Zubrinich C, Marshall J, Cook D. The “Consent to Research” paradigm in critical care: challenges and potential solutions. Intensive Care Med 2009;35:1655–1658. |
| 10. | Holt R, Siddiqi N, Young J. The ethics of consent in delirium studies. J Psychosom Res 2008;65:283–287. |
| 11. | Annane D, Sébille V, Charpentier C, Bollaert PE, François B, Korach JM, Capellier G, Cohen Y, Azoulay E, Troché G, et al. Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock. JAMA 2002;288:862–871. |
| 12. | Harvey S, Harrison DA, Singer M, Ashcroft J, Jones CM, Elbourne D, Brampton W, Williams D, Young D, Rowan K; PAC-Man Study Collaboration. Assessment of the clinical effectiveness of pulmonary artery catheters in management of patients in intensive care (PAC-Man): a randomised controlled trial. Lancet 2005;366:472–477. |
| 13. | Roberts I, Yates D, Sandercock P, Farrell B, Wasserberg J, Lomas G, Cottingham R, Svoboda P, Brayley N, Mazairac G, et al.; CRASH Trial Collaborators. Effect of intravenous corticosteroids on death within 14 days in 10008 adults with clinically significant head injury (MRC CRASH Trial): randomised placebo-controlled trial. Lancet 2004;364:1321–1328. |
| 14. | Perner A, Ibsen M, Bonde J. Attitudes to drug trials among relatives of unconscious intensive care patients. BMC. Anesthesiology 2010;10:6. |
| 15. | Burns KEA, Magyarody NM, Duffett M, Nisenbaum R, Cook DJ. Attitudes of the general public toward alternative consent models. Am J Crit Care 2011;20:75–83. |
| 16. | Duffett M, Burns K, Kho ME, Lauzier F, Meade MO, et al. Consent in critical care trials: a survey of Canadian research ethics boards and critical care researchers. J Crit Care 2011;26:533.e11–533.e22. |
| 17. | Mitka M. Aiding emergency research aim of report on exceptions to informed consent. JAMA 2007;298:2608–2609. |
| 18. | Burns KEA, Raptis S, Rizvi L, Cook DJ. Coenrollment in a national cross-sectional study of research recruitment practices: the Consent Study [abstract]. Am J Respir Crit Care Med 2012;185:A2544. |
| 19. | Burns KEA, Zubrinich C, Tan W, Smith O, McDonald E, Skrobik Y, Saginur R, Jeffs L, Marshall J, Heslegrave R, et al. Understanding barriers to recruitment and the rationale for providing consent for participation in critical care research: the Consent Study (a CCCTG project) [abstract]. Am J Respir Crit Care Med 181;2010:A6047. |
| 20. | Wilkins R. PCCF + version 5F user’s guide: automated geographic coding based on the Statistics Canada postal code conversion files. Catalogue no. 82F0086-XDB. Ottawa, Ontario, Canada: Health Analysis Division, Statistics Canada; 2010. |
| 21. | Kleinbaum DG. Applied regression analysis and other multivariable methods. Boston: Duxbury Press; 1998, pp. xviii, 798. |
| 22. | Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007;165:710–718. |
| 23. | Grap MJ, Munro CL. Subject recruitment in critical care nursing research: a complex task in a complex environment. Heart Lung 2003;32:162–168. |
| 24. | Glassberg AE, Luce JM, Matthay MA; National Heart, Lung, and Blood Institute Clinical Trials Network. Reasons for nonenrollment in a clinical trial of acute lung injury. Chest 2008;134:719–723. |
| 25. | Marshall J, Cook DJ. Barriers to investigator-initiated research: a survey of the Canadian Critical Care Trials Group [abstract]. Am J Respir Crit Care Med 2005;33:A82. |
| 26. | Bellomo R, Cass A, Cole L, Finfer S, Gallagher M, Goldsmith D, Myburgh J, Norton R, Scheinkestel C; RENAL Replacement Therapy Trial Investigators. Screening and study enrolment in the Randomized Evaluation of Normal vs. Augmented Level (RENAL) Replacement Therapy Trial. Blood Purif 2009;27:199–205. |
| 27. | Crowley ST, Chertow GM, Vitale J, O’Connor T, Zhang J, Schein RM, Choudhury D, Finkel K, Vijayan A, Paganini E, et al.; VA/NIH Acute Renal Failure Trial Network Study Group. Lessons for successful study enrollment from the Veterans Affairs/National Institutes of Health Acute Renal Failure Trial Network Study. Clin J Am Soc Nephrol 2008;3:955–961. |
| 28. | Mancebo J, Fernández R, Blanch L, Rialp G, Gordo F, Ferrer M, Rodríguez F, Garro P, Ricart P, Vallverdú I, et al. A multicenter trial of prolonged prone ventilation in severe acute respiratory distress syndrome. Am J Respir Crit Care Med 2006;173:1233–1239. |
| 29. | Robertson CS, McCullough LB, Brody B. Finding family for prospective consent in emergency research. Clin Trials 2007;4:631–637. |
| 30. | Saver JL, Kidwell C, Eckstein M, Ovbiagele B, Starkman S; FAST-MAG Pilot Trial Investigators. Physician-investigator phone elicitation of consent in the field: a novel method to obtain explicit informed consent for prehospital clinical research. Prehosp Emerg Care 2006;10:182–185. |
| 31. | Cook DJ, Blythe D, Rischbieth A, Hebert PC, Zytaruk N, Menon K, Erikson S, Fowler R, Heels-Ansdell D, Meade MO. Enrollment of intensive care unit patients into clinical studies: a trinational survey of researchers’ experiences, beliefs, and practices. Crit Care Med 2008;36:2100–2105. |
| 32. | Cook D, McDonald E, Smith O, Zytaruk N, Heels-Ansdell D, Watpool I, McArdle T, Matte A, Clarke F, Vallance S, et al.; PROTECT Research Coordinators; PROTECT Investigators; Canadian Critical Care Trials Group; Australian and New Zealand Intensive Care Society Clinical Trials Group. Co-enrollment of critically ill patients into multiple studies: patterns, predictors and consequences. Crit Care 2013;17:R1. |
| 33. | Soltner C, Lassalle V, Galienne-Bouygues S, Pottecher J, Floccard B, Delapierre L, Jungfer F, Sargentini C, Alberti C, Beydon L; Lifrea Group. Written information that relatives of adult intensive care unit patients would like to receive—a comparison to published recommendations and opinion of staff members. Crit Care Med 2009;37:2197–2202. |
| 34. | Fan E, Shahid S, Kondreddi VP, Bienvenu OJ, Mendez-Tellez PA, Pronovost PJ, Needham DM. Informed consent in the critically ill: a two-step approach incorporating delirium screening. Crit Care Med 2008;36:94–99. |
| 35. | Egerod I, Christensen BV, Johansen L. Trends in sedation practices in Danish intensive care units in 2003: a national survey. Intensive Care Med 2006;32:60–66. |
| 36. | Mehta S, Burry L, Fischer S, Martinez-Motta JC, Hallett D, Bowman D, Wong C, Meade MO, Stewart TE, Cook DJ; Canadian Critical Care Trials Group. Canadian survey of the use of sedatives, analgesics, and neuromuscular blocking agents in critically ill patients. Crit Care Med 2006;34:374–380. |
| 37. | Chertow GM, Pascual MT, Soroko S, Savage BR, Himmelfarb J, Ikizler TA, Paganini EP, Mehta RL; PICARD. Reasons for non-enrollment in a cohort study of ARF: the Program to Improve Care in Acute Renal Disease (PICARD) experience and implications for a clinical trials network. Am J Kidney Dis 2003;42:507–512. |
| 38. | Scales DC, Smith OM, Pinto R, Barrett KA, Friedrich JO, Lazar NM, Cook DJ, Ferguson ND. Patients’ preferences for enrolment into critical-care trials. Intensive Care Med 2009;35:1703–1712. |
| 39. | Barrett K, Ferguson ND, Athaide N, Cook DJ, Friedrich JO, McDonald E, Pinto R, Smith OM, Scales DC. Substitute decision makers' attitudes towards research decision-making for critically ill patients. Int Care Med 2012;38:1616–1623. |
| 40. | Biros MH, Lewis RJ, Olson CM, Runge JW, Cummins RO, Fost N. Informed consent in emergency research: consensus statement from the coalition conference of acute resuscitation and critical care researchers. JAMA 1995;273:1283–1287. |
| 41. | Pinder M, Tshukutsoane S, Scribante J, Piccolo R, Lipman J. Critical care research and pre-emptive informed consent: a practical approach used in Chris Hani Baragwanath ICU. Intensive Care Med 1998;24:353–357. |
| 42. | Aramendi JI, Mestres C-A, Martinez-León J, Campos V, Muñoz G, Navas C. Triflusal versus oral anticoagulation for primary prevention of thromboembolism after bioprosthetic valve replacement (TRAC): prospective, randomized, co-operative trial. Eur J Cardiothorac Surg 2005;27:854–860. [Published erratum appears in Eur J Cardiothorac Surg 28:188] |
| 43. | Ciroldi M, Cariou A, Adrie C, Annane D, Castelain V, Cohen Y, Delahaye A, Joly LM, Galliot R, Garrouste-Orgeas M, et al.; Famirea Study Group. Ability of family members to predict patient’s consent to critical care research. Intensive Care Med 2007;33:807–813. |
| 44. | Mehta S, Quittnat Pelletier F, Brown M, Ethier C, Wells D, Burry L, MacDonald R. Why substitute decision makers provide or decline consent for ICU research studies: a questionnaire study. Intensive Care Med 2012;38:47–54. |
| 45. | Luce JM, Cook DJ, Martin TR, Angus DC, Boushey HA, Curtis JR, Heffner JE, Lanken PN, Levy MM, Polite PY, et al.; American Thoracic Society. The ethical conduct of clinical research involving critically ill patients in the United States and Canada: principles and recommendations. Am J Respir Crit Care Med 2004;170:1375–1384. |
| 46. | Barrett KA, Scales DC. Considering the vulnerabilities of surrogate decision-makers when obtaining consent for critical care research. Intensive Care Med 2012;38:4–6. |
Supported by a Catalyst Grant in Ethics from the Canadian Institutes of Health Research (CIHR). The funding agency had no role in the study design, data collection and interpretation, reporting, or the results and the decision to submit the manuscript for publication. The researchers are independent from the study funders; however, K.E.A.B. holds a Phase 2 Clinician Scientist award from the CIHR and D.J.C. is a Research Chair with the CIHR.
Author Contributions: K.E.A.B., O.S., E.M., J.C.M., R.S., R.H., G.R., and D.J.C. developed the study protocol and secured peer-reviewed funding to conduct the study. K.E.A.B. and C.Z. organized research ethics approvals and supervised study implementation and data collection. K.E.A.B., C.Z., W.T., S.R., O.S., J.C.M., and D.J.C. organized data entry, resolved data queries, and prepared the database for statistical analysis. K.E.A.B., O.S., E.M., J.C.M., R.S., R.H., G.R., and D.J.C. prioritized variables for inclusion in the multivariate analysis. W.X. conducted the planned analyses. All authors participated in drafting and revising the manuscript for important intellectual content. All authors approved the final version of the manuscript.
This article has an online supplement, which is available from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201208-1537OC on March 22, 2013
Author disclosures are available with the text of this article at www.atsjournals.org.
