A Qualitative Analysis of Factors Influencing Critical Care Trial Enrollment Among Surrogates

Abstract

Background:

We sought to identify factors that influence surrogate decision makers’ decisions to enroll patients into a critical care randomized controlled trial.

Methods:

We conducted a qualitative study embedded within a randomized controlled trial testing the effect of a behavioral nudge intervention for surrogate decision makers on enrollment rate in a sham ventilatory weaning trial among patients with acute respiratory failure. Participants were adult surrogate decision makers of patients receiving mechanical ventilation for acute respiratory failure. The study was conducted in 10 ICUs across 2 urban hospitals within an academic medical center in Philadelphia, Pennsylvanaia, United States. Immediately following their trial enrollment decision, surrogate decision makers were asked to enter free-text responses about the factors that influenced their decision. Responses were analyzed using content analysis.

Results:

Of the 90 (49%) participants who provided free-text responses, the mean age was 54.9 years (SD 14.3), 69 (79%) were Caucasian, and 48 (53%) were the spouse of the eligible patient. We identified 5 themes influencing enrollment decisions: (i) trial characteristics, (ii) patient clinical condition, (iii) decision making processes, (iv) altruism, and (v) enrollment attempt. Among surrogates who enrolled the patient in the trial (n = 40), the most commonly cited factors were helping future patients (n = 24, 60%) and following the patient’s wishes (n = 11, 28%). In contrast, those who declined enrollment (n = 50) most commonly reported that the patient was too sick (n = 27, 54%) and that they feared complicating the patient’s condition (n = 11, 22%).

Conclusions:

Surrogates who enroll patients into trials most often cite altruistic motivations, while those who decline enrollment are most often concerned with the severity of the patients’ condition.

Background

One of the most challenging aspects of conducting randomized controlled trials (RCTs) is participant recruitment, particularly in the intensive care unit (ICU) where the majority of enrollment decisions are made by surrogate decision makers (SDMs). The SDM role is already cognitively and emotionally burdensome, and research enrollment decisions may contribute to that burden and lead to post-traumatic stress symptoms among SDMs.¹

Previous studies that have attempted to describe factors that influence SDMs’ decisions whether or not to enroll a critically ill patient into research studies are limited by recall bias due to delay between the trial enrollment attempt and follow-up.^2,3 Further, these studies provided insufficient information about the parent studies, which may provide important context.

We embedded a qualitative study within a RCT testing the effect of a behavioral nudge intervention for SDMs on enrollment rate in a sham ventilatory weaning trial among patients with acute respiratory failure (ARF).⁴ Immediately following their trial enrollment decision, SDMs were queried about the influential factors in their decision-making process. This work will inform the development of SDM-centered recruitment practices and novel interventions to overcome modifiable barriers to critical care trial participation.

Methods

Setting and Participants

Details of the parent RCT for this study were previously reported.⁴ Briefly, we conducted an RCT of a pre-consent nudge survey to increase enrollment into a sham ventilator weaning trial among 182 SDMs of critically ill patients. The study was conducted in ten medical, surgical, or mixed ICUs across 2 urban hospitals within the University of Pennsylvania Health System. Eligible participants were English-proficient adult (>18 years) SDMs of adult patients in the ICU for >24 hours with ARF and receiving mechanical ventilation. SDMs were excluded if extubation was anticipated within 24 hours. Additionally, due to the sham nature of the trial, involving temporary deception, out of an abundance of caution SDMs were excluded if the treating clinician perceived strain in the relationship with or general healthcare distrust by the SDM. The University of Pennsylvania Institutional Review Board approved this study with a waiver of informed consent.

Data Collection

Following the enrollment decision, SDMs completed a demographic questionnaire about the patient and themselves. We then asked SDMs “What factors influenced your decision to participate or not to participate in this trial?” with a free-text box for response. All data were collected in-person, in the patient’s room or in a conference room, via the Research Electronic Data Capture (REDCap) platform.⁵ Data were collected by D.C.K, a pulmonary and critical care fellow enrolled in a masters of science in clinical epidemiology program and K.L.O, an experienced research assistant enrolled in a masters of science in nursing program. Interviewers introduced themselves as “ICU researchers” and neither had a previously established relationship with participants. Immediately following data collection, research coordinators engaged all SDMs in a standardized debriefing process to disclose the sham trial. Participants were not provided compensation for participation.

Data Analysis

We used content analysis to analyze SDMs’ free-text responses.⁶ Two investigators (D.C.K. & K.R.C.) independently open coded SDMs’ responses, and generated a list of factors that emerged from each response. Each unique factor was recorded only once per response. The factors were assessed for similarities and organized into themes to develop a codebook. Two investigators (D.C.K. and B.D.H.) then independently coded all responses using the codebook, revising and recoding iteratively;⁷ discrepancies between reviewers were resolved by arbitration (K.R.C.).

We report the frequency of factors by study arm, by enrollment decision and by prior surrogate research experience. All analyses were performed using Stata v14.2 (StataCorp, College Station, TX).

Results

Participant Characteristics

Ninety (49%) SDMs in the parent RCT provided a response to the free-text study question. The mean age of respondents was 54.9 years (SD 14.3), 69 (79%) were Caucasian, and 48 (53%) were the spouse of the eligible patient (Table 1). Fifty (56%) participants reported knowing someone who had participated in medical research, and 21 (24%) had participated in research themselves.

Table 1.

Characteristics of Participants Who Provided Responses.

Characteristic	No. (%)
Total no. providing data	90
Surrogate age, mean (SD)	54.9 (14.3)
Surrogate female sex	69 (77%)
Surrogate racea
White or Caucasian	69 (79%)
Black or African-American	11 (13%)
Other	7 (8%)
Surrogate ethnicity, Hispanica	3 (3%)
Surrogate educationa
High school graduate	16 (18%)
Some college	25 (28%)
College graduate	27 (30%)
Advanced degree	21 (24%)
Surrogate relationship to patient
Spouse or partner	48 (53%)
Child	18 (20%)
Parent	13 (14%)
Other	11 (12%)
Surrogate prior ICU admission	12 (13%)
Surrogate prior research participationa	21 (24%)
Surrogate prior research exposurea,b	50 (56%)
Patient primary diagnosis
Abdominal surgery	7 (8%)
Cardiac disease	8 (9%)
Cardiac surgery	12 (13%)
Malignancy	2 (2%)
Neurologic disease	11 (12%)
Other medical	8 (9%)
Other surgery	4 (4%)
Pulmonary infection	12 (13%)
Respiratory failure	7 (8%)
Septic shock	12 (13%)
Transplant	2 (2%)
Trauma	5 (6%)
Patient APACHE Score, mean (SD)	102.3 (34.3)

^aMissing data: race (n = 3), ethnicity (n = 3), education (n = 1), prior research participation (n = 1), prior research exposure (n = 1).

^b“Has anyone you have known ever participated in a medical research trial?”.

Factors Impacting Enrollment Decision

We identified 5 themes comprising 20 factors: (i) trial characteristics, (ii) patient clinical condition, (iii) decision-making processes, (iv) altruism, and (v) enrollment attempt (Table 2). Each SDM response was a median 1 (IQR 1–1) sentence long and included a median of 2 (IQR 1–2) factors. Among SDMs who decided to enroll the patient in the RCT (n = 40), the most commonly cited factors were helping future patients (n = 24, 60%), following the patient’s wishes (n = 11, 28%), having favorable views of biomedical research (n = 9, 23%), and the perceived safety of the trial (n = 9, 23%). In contrast, those who declined enrollment (n = 50) most commonly reported that the patient was too sick (n = 27, 54%), fear of “rocking the boat” and complicating the patient’s condition (n = 11, 22%), and potential interference with the medical team’s decisions (n = 10, 20%). The frequency of reported factors did not substantially differ between intervention and control groups in the parent RCT (Table 2) or by prior surrogate research experience (Online Supplement, Appendix C).

Table 2.

Factors and Themes Identified in Surrogate Responses.

			No. (%) surrogate decision makers
			Enrollment decision		Trial arm
Themes	Factors	Exemplar quotes	Declined (n = 50)	Enrolled (n = 40)	Intervention (n = 42)	Control (n = 48)
Study characteristics	Perception of risk	“We felt that there wasn’t any risk involved.”	2 (4)	9 (23)	4 (10)	7 (15)
	Interference with clinical decisions	“I want to have the critical care team have full accountability and responsibility for decisions with no outside influence.”	10 (20)	0	4 (10)	6 (13)
	Clinical team override	“If the ICU team can override the protocol then there isn’t much risk.”	0	6 (15)	2 (5)	4 (8)
	Randomization	“I’m just a little nervous about what protocol would be selected and if this would be good for my husband.”	3 (6)	1 (3)	3 (7)	1 (2)
	Perception of benefit	“Love for a family member that needs to progress as quickly as possible for the next step in his care.”	0	1 (3)	1 (2)	0
Patient’s clinical condition	“Rocking the boat”	“Her health is too delicate to add another factor to the equation.”	1 1 (22)	0	6 (14)	5 (10)
	Acute condition	“Because the patient is critically unstable.”	27 (54)	1 (3)	14 (33)	14 (29)
	Complexity	“Her complicated condition, with regards to her lungs, makes me uneasy.”	5 (10)	0	2 (5)	3 (6)
	Medical history or comorbidities	“My loved one’s comorbidities.”	3 (6)	0	1 (2)	2 (4)
Decision-making process	Patient’s wishes	“My wife has always supported using her unfortunate condition to potentially help someone in the future who may benefit from her experiences. 1 believe she would want to participate in this experiment.”	3(6)	1 1 (28)	7 (17)	7 (15)
	Personal views of research	“We believe in research to help advance treatment.”	3 (6)	9 (23)	4 (10)	8 (17)
	Surrogate stress or distress	“I’m not emotionally sound to make this decision presently... my husband is critically ill!”	5 (10)	0	1 (2)	4 (8)
	Collective decision	“Our family talked about it and overall decided this wasn’t for us.”	3 (6)	0	0	3 (6)
	Difficult decision	“I’m his niece and feel like 1 shouldn’t make that decision for him.”	2 (4)	0	1 (2)	1 (2)
	Surrogate’s medical knowledge	“Patient has been unsuccessful on pressure ventilation and 1 believe treatment goal is to reduce FiO2 to ween so 1 would be concerned if she was assigned to the pressure arm.”	1 (2)	1 (3)	1 (2)	1 (2)
Altruism	Helping future patients	“My mother would want to help future patients in any way possible.”	0	24 (60)	1 1 (26)	13 (27)
	Paying back healthcare system	“This hospital is doing a valiant effort of trying to bring my husband back to health, so the least we can do is to help you by participating in a study that can help others.”	0	3 (8)	0	3 (6)
Enrollment attempt	Timing of approach	“We are waiting to hear more about care from the clinical team.”	4(8)	0	1 (2)	3 (6)
	Delivery of information	“The study was explained fully and is very straight forward.”	0	4 (10)	2 (5)	2 (4)
	Clinical team supports enrollment	“The medical team’s opinion.”	0	2 (5)	0	2 (4)

Discussion

We identified 5 themes of factors that influence SDMs considering whether or not to enroll a patient in a critical care trial. Factors that influenced SDMs who elected to enroll the patient differed from those who declined. Understanding what motivates SDMs to enroll patients in critical care trials can inform trial recruitment practices and lead to interventions that target common concerns or misperceptions.

Consistent with previous studies of ICU SDMs’ trial enrollment decisions, we found that the decision-making process is complex and rarely conforms to pure substitutive decision making. Two previous studies evaluated factors important to ICU SDMs who had been approached for patient enrollment into an observational study or randomized trial. Both studies identified a desire to hasten recovery as the most important factor in SDMs’ enrollment decisions.^2,3 This has been termed therapeutic misconception—i.e., conflation of research with clinical care⁸—and is generally regarded as an ethically inappropriate reason for enrolling in research. In contrast, only 1 SDM in our cohort anticipated a direct benefit of the research to the patient. It is possible that because the parent trial for this study involved a comparison of 2 standard approaches to ventilator weaning, there was not an implied direct benefit beyond usual practices.

While few SDMs who declined enrollment reported the actual trial participation risks discussed during informed consent, many expressed a concern that the patient was “too sick” or that introducing the additional uncertainty of trial participation may “rock the boat”. Indeed, surrogate perception of the riskiness of trial participation, along with surrogate race, were found to be independently associated with trial enrollment.⁴ Considering participation in the sham trial conferred minimal risk beyond usual care, these concerns more likely reflected SDMs’ emotions about the patient’s condition rather than an objective evaluation of risks of trial participation. These findings are consistent with barriers identified by Mehta et al, including concern about patient harm and being too anxious to consider research.² Indeed, psychological burdens among ICU SDMs are common, and have been shown to be independently associated with research participation decisions.¹ The relationship between SDMs’ anxiety about the patient’s condition and decision making about participation in research warrants further investigation.

Our results suggest that altruism was a powerful motivator for trial enrollment, consistent with findings by Mehta et al. The pre-consent nudge survey used in the parent RCT was designed to highlight the altruistic nature of research participation, though it was not found to be effective in increasing the enrollment rate.⁴ However, alternative methods of encouraging altruism and personalizing a duty of reciprocity wherein the surrogate recognizes that the patient is benefiting from previous patients’ participation in research and thus feels an obligation to participate may be more effective. For example, researchers could highlight how patient participation in past ICU trials have contributed to improvements in outcomes from which the patient may now be benefiting. Additionally, investigators may try norm setting by providing a general description of other trial participants’ clinical conditions to counteract SDM’s misperception that the patient is “too sick” to participate.⁹ Lastly, although the ability of the ICU team to override the research protocol for any reason at any time was discussed in the informed consent, some SDMs declined enrollment because they were concerned that the patient’s treatment decisions would be influenced by the trial. Thus, future recruitment efforts should include clear assurances that the ICU team remains in charge of the patient’s care regardless of the trial protocol.

Strengths of this study include the use of a sham trial to mimic real-world trial enrollment decisions and the factors that influenced them, since hypothetical scenarios poorly reflect actual research participation decisions.¹⁰ Additionally, SDMs represented patients with diverse critical illness conditions being cared for by clinicians with various specialty backgrounds in ten different ICUs across 2 hospitals. Finally, SDMs were questioned immediately following the enrollment decision thus minimizing recall bias, which is exacerbated in the ICU setting where SDMs experience tremendous clinical decision-making burdens.²

This study has several limitations to consider. First, generalizability may be limited as this study was conducted within a single healthcare system and involved a single trial intervention. Factors influencing SDMs’ trial enrollment decision may differ depending on the nature of the intervention being studied. While the perception of risk⁴ and frequency of altruistic motivations were similar between those exposed to the nudge and those unexposed (Table 2), the nudge may have influenced decision making in unmeasured ways and thus may impact generalizability. Second, the majority of SDMs who declined to answer the study question also declined to provide demographic information, thus precluding assessment for nonresponse bias. Along with our decision to exclude SDMs with perceived distrust in healthcare from the parent trial, this suggests that there may be additional influential factors not elucidated in this study cohort. Third, although the introduction, initial description of the trial, and responses to frequently asked questions were scripted, there was variability in the discussion that followed with each surrogate which may have influenced the results and therefore generalizability. Finally, participants responded to one open-ended question with brief responses, thus additional important contextual information may be uncovered through qualitative interviews.

Conclusion

We identified differences in the factors that influenced SDMs to enroll a patient in a critical care trial compared to those who declined participation. Altruism was the strongest motivator of SDMs’ decision to enroll the patient, while SDMs’ perception of the severity of the patient’s condition often influenced them to decline enrollment. Designing research recruitment strategies that address modifiable barriers among SDMs may improve critical care trial enrollment.