Improving Research Enrollment of Severe Burn Patients

Abstract

Objective

Enrolling severely burn injured patients into prospective research studies poses specific challenges to investigators.

Methods

We describe our experience of recruiting adults with ≥20% total body surface area burns or inhalation injury admitted to a single academic burn unit into observational research with minimally invasive specimen collection. We outline iterative changes that we made to our recruitment processes in response to perceived weaknesses leading to delays in enrollment. Our primary outcome was the change in days to consent for enrolled patients or cessation of recruitment for non-enrolled patients before and after the interventional modifications. We assessed change in overall enrollment as a secondary outcome.

Results

Study enrollment was ~70% in both 4-month study periods before and after the intervention. Following the intervention, time to consent by surrogate decision maker decreased from a median of 26.5 days (IQR 14–41) to 3 days (IRQ 3–6) (p=0.004). Time to initial consent by patient changed from a median of 15 days (IQR 2–30) to 3 days (IQR 2–6) (p=0.27). Time to decline for non-enrolled patients decreased from a median of 12 days (IQR 6.5–27) to 1.5 days (IQR 1–3.5) (p=0.026).

Conclusions

Both the findings of our study and our brief literature review suggest that careful design of the recruitment protocol, increased experience of the study team, and broad time windows for both approach and enrollment improve the efficiency of recruiting critically ill burn patients into research.

Introduction

Each year in the United States, 486,000 people receive medical treatment for burn injuries and 40,000 of these are hospitalized, including 30,000 who are hospitalized at burn centers.¹ Efforts to improve outcomes from burn injuries rely on a better understanding of the factors related to the injury including the pathophysiology, barriers to treatment, and factors that increase morbidity and mortality. Enrollment of burn patients is usually straight-forward in the case of observational studies and repositories for which informed consent may not be required.² However, less is known about issues surrounding enrollment of these patients in prospective studies with informed consent. Prospective studies improve data quality compared to retrospectively collected data, and improving enrollment into prospective studies allows data to be obtained more efficiently, thereby increasing the generalizability of the data collected.

Several studies have been published in the last decade regarding the recruitment of critically ill patients,³ patient preferences for enrollment in critical-care trials,⁴ and the attitudes of surrogate decision makers’ (SDM) towards research decision making for critically ill patients.⁵ However, burn injured patients have long been acknowledged to be a unique population. For example, severely burned patients were excluded from the early Acute Respiratory Distress Syndrome Network studies.⁶ Dr. Saffle, a well-recognized surgeon, has also described the social and medical challenges involved in the care of burn patients that is different from other patients.⁷ Therefore, findings from other populations of critically ill patients may not be applicable to burn injured patients.

In this paper, we compared time to enrollment before and after iterative changes to the process of recruiting severely burned patients into two non-interventional studies at a busy academic burn center (>1500 burn unit admissions annually). We incorporate our findings into a brief literature review of the barriers, motivators, and roles of SDMs and physicians in the recruitment of patients into acute care studies.

Methods

Study design

Our institution is an 850-bed academic tertiary care facility including a 21-bed intensive care unit (ICU) for severely ill adult and pediatric patients with burns or extensive exfoliating skin conditions. In November 2015, we began enrolling patients into two non-interventional studies: 1) an observational study of infections occurring during burn ICU admission with analysis of Gram-negative bacterial isolates obtained from clinical patient specimens, and 2) a minimally invasive study with collection of weekly surveillance wound and peri-rectal cultures using sterile swabs and twice weekly aseptically collected tracheal aspirates for intubated patients with for culture to isolate Gram-negative bacteria from these samples. The goal of these studies were to identify changes in the microbial milieu of patients admitted to the burn unit during their hospital stay. We included patients who met the following criteria 1) age ≥18 years, 2) percentage total body surface area burn ≥20 or inhalation injury, 3) fluency in either English or Spanish, 4) presentation within 2 days of injury, and 5) anticipated length of stay in the burn ICU of at least 5 days.

Consent and recruitment

Both studies were approved by our Institutional Review Board (IRB) and operated with informed consent. Initial consent from a SDM was acceptable for patients unable to provide consent. Non-consented patients were given the opportunity to withdraw study consent provided by a SDM when they were no longer incapacitated or impaired.

The principal investigator (PI), a regular consulting physician in the burn unit but not a burn surgeon, was involved in the recruitment process for some patients. Formal informed consent was obtained by a trained research coordinator. Multiple strategies were instituted prior to study commencement to maximize recruitment. These strategies are listed in Table 1 and included, for example, the availability of both English and Spanish consent forms.

Table 1.

Strategies to maximize enrollment of severely injured burn patients.

Pre-study strategies

Mid-study interventional modifications

1. Consents forms in English and Spanish 2. Phone scripts and IRB approval to obtain verbal consent from SDM’s by phone 3. IRB approval to obtain verbal consent for patients unable to sign forms due to burn injuries 4. Option for observational participation only or option to participate with only some invasive samples 5. IRB approval to obtain admission samples on intubated patients with a waiver of consent 6. Involvement of caseworker to assist in identifying the SDM 7. Research assistant availability during somewhat flexible hours for recruitment 8. No strictly defined time window for enrollment

1. Deliberate delay of approach of SDM and patients until ~48 hours after admission and until pain controlled 2. Physician PI outreach if no response obtained from SDM after 48 hours of approach by research coordinator 3. Request participation in the observational study prior to discussing the study with minimally invasive sample collection 4. Mail or email consent forms to SDM after obtaining verbal phone consent with phone script 5. More frequent communication between the physician PI and the consenting research coordinator

IRB, institutional review board; SDM, surrogate decision maker; PI, principle investigator

Iterative changes in the recruitment process

Throughout the initial four months of the study (Nov 9, 2015–Feb 29, 2016), we perceived several weaknesses in our recruitment process leading to delays in enrollment. As a result, several iterative modifications were made to our recruitment process during months 4–6 of the study to try to improve this process (Table 1). First, patients and SDMs appeared to decline a discussion of research on admission due to pain or feeling overwhelmed. Therefore, we delayed the approach of SDMs until 48 hours after admission and delayed approaching patients by at least 24–48 hours and until pain was adequately controlled. We began to recruit patients to the study that did not require invasive sample collection first; if this study was deemed to be potentially acceptable, we then proceeded to broach the second study involving patient specimen collection. Prior to study initiation, the IRB approved a waiver of written consent for physically injured patients and permission to obtain verbal consent from SDMs with phone scripts. However, in the early study period, we often had delays waiting for the SDM to receive and read physical copies of study forms sent by mail or email before attempting to obtain consent. So in later months, we mailed consent forms after consent was obtained by phone and offered the SDM the opportunity to withdraw consent if the SDM felt that paper documents were incongruent with the phone script. As some patients and SDMs appeared to respond better to approach by the PI, the PI became increasingly involved in the recruitment process throughout the later months. Finally, the PI and research coordinator also standardized the processes of screening, approaching patients, and communicating with each other to limit duplicate work and delays. We used the second four months of the study (Mar 1, 2016–June 24, 2016) to represent time after our interventional modifications.

Data analysis

Results are displayed as median and interquartile range (IQR) for continuous variables and numbers and percentages for categorical variables. Bivariate analyses were performed to examine the associations between patient characteristics and enrollment as well as the associations between enrollment and time to enrollment in the first four months (before intervention) and second four months of the study (after intervention). As several continuous variables were non-normal and the sample size was small, the Wilcoxon rank sum test was used for continuous variables and the Fisher exact test was used for categorical variables. All P values were 2-sided, and P<0.05 was considered statistically significant. Stata statistical software (release 13; StataCorp LP, College Station, Texas) was used for all analyses.

The two extracorporeal membrane oxygenation (ECMO) patients transferred units before a SDM could be approached and were considered non-enrolled as of the day of transfer as they were no longer eligible for recruitment at that time. One ECMO patient was eligible for the observational study alone after returning to burn unit following ECMO: he was consented by SDM but later withdrew consent. To be conservative, he remained as non-enrolled as of the day of transfer in our analyses.

Results

Over an 8 month period from November 9, 2015 through June 24, 2016, 296 admissions to our burn ICU were screened for eligibility (Figure 1). Forty patients met inclusion criteria. Of eligible patients, 27 (68%) were male, 28 (70%) were white, and 40 (100%) spoke English as their preferred language. Twenty-eight patients (70% of those eligible) were enrolled, 3 (7.5%) patients declined, 3 (7.5%) SDMs declined, 4 (10%) died prior to recruitment, and 2 (5%) were transferred out of the burn ICU for ECMO during recruitment. Of eligible patients, thirty-four patients (85%) required initial recruitment via a SDM, while 6 (15%) patients were recruited directly. Four patients (10%) began the recruitment process requiring a SDM but regained the ability to consent for themselves before the SDM could be identified or before the SDM made a decision regarding study participation. One patient withdrew consent after it was provided by a SDM. One patient directly consented to the observational study alone, but all other patients or SDMs consented to or declined both studies.

Figure 1.

Study flow diagram. ICU, intensive care unit; TBSA total body surface area; SDM, surrogate decision maker; ECMO, extracorporeal membrane oxgenation

Characteristics of all patients who were and were not enrolled in either study during the entire study period are shown in Table 2. Compared to those patients who were not enrolled, enrolled patients were more often male (71% vs. 58%), more often white (79% vs. 50%), and more often had inhalation injury (50% vs. 33%). However, none of these characteristics reached statistical significance in our small study population.

Table 2.

Characteristics of all patients eligible for study enrollment^*

	Eligible patients enrolled in either study N = 28 (70%)	Eligible patients not enrolled N = 12 (30%)	P†
Age at admission years, median, IQR	51.5 (36–63)	56.5 (40–69)	0.36
Sex
Male	20 (71)	7 (58)	0.48
Female	8   (29)	5 (42)
Race
White	22 (79)	6 (50)	0.07
Black	6   (21)	4 (33)
Other/Unknown	0   (0)	2 (16)
Preferred language
English	28 (100)	12 (100)	NA
Spanish	0   (0)	0   (0)
Total body surface area percentage, median, IQR	18.0 (1–37.5)	16.5 (0–25)	0.32
Inhalation injury
Yes	14 (50)	4 (33)	0.49
No	14 (50)	8 (67)
Decision-maker
Surrogate	21 (75)	3 (25)	0.33‡
Patient	7   (25)	3 (25)
Not applicable (died/transferred prior to recruitment)	0   (0)	6 (50)

IQR, interquartile range; NA, not applicable.

^*Includes all patients pre- and post-intervention. Data are given as number (percentage) unless otherwise indicated.

^†P values were calculated with the Wilcoxon rank sum test for continuous variables and the Fisher exact test for categorical variables.

^‡Patients who died/transferred prior to recruitment were excluded when calculating the p value for decision-maker.

Study enrollment prior to the intervention was 68% and 73% (p=1.0) in the later time period after the intervention (Table 3). Following the intervention, time to consent by a SDM decreased from a median of 26.5 days (IQR 14–41) to 3 days (IRQ 3–6). Time to initial consent by patients changed from a median of 15 days (IQR 2–30) to 3 days (IQR 2–6). Time to decline of non-enrolled patients decreased from a median of 12 days (IQR 6.5–27) to 1.5 days (IQR 1–3.5). Both the change in time to consent by SDM and the change time to cessation of recruitment were statistically significant (p<0.05).

Table 3.

Differences in enrollment before and after the intervention.

	Prior to intervention Nov 9, 2015–Feb 29, 2016 (112 days)	After intervention Mar 1, 2016–June 24, 2016 (115 days)	P*
Screened patients, N	139	157
Eligible patients, N (% of patients screened)	25 (18)	15 (10)
Enrollment, N (% of patients eligible)
Enrolled patients	17 (68)	11 (73)	1.0
Initial consent by SDM, N (% enrolled)	14 (82)	7 (64)
Initial consent by self, N (% enrolled)	3 (18)	4 (36)
Non-enrolled patients†	8 (32)	4 (27)
Days to initial consent or decline, median, IQR
Days to initial consent by SDM	26.5 (14–41)	3 (3–6)	0.004
Days to initial consent by self	15 (2–30)	3 (2–6)	0.27
Days to decline for non-enrolled patients	12 (6.5–27)	1.5 (1–3.5)	0.026

SDM, surrogate decision maker; IQR, interquartile range.

^*P values were calculated with the Wilcoxon rank sum test for continuous variables and the Fisher exact test for categorical variables.

^†Patient/SDM declined or patient died/transferred during recruitment

Discussion

An insufficient amount of literature exists to advise investigators on how to conduct clinical trials in burn patients, specifically addressing the barriers to research participation and how to overcome those barriers to improve recruitment into burn studies. Despite a study design that anticipated several barriers in the pre-study protocol design phase, we struggled to enroll patients in a timely and efficient manner. We made several changes to our recruitment process that decreased the time to patient enrollment or decline. Although the downstream consequences of our recruitment changes were not measured, our interventions likely decreased wasted effort by research coordinators on missed SDM and patient encounters, decreased delays to patient chart abstraction, and improved time to collection of the second surveillance specimen.

Compared to other critically ill populations, several advantages exist that would seem to favor recruitment of severely injured burn patients into research. Although types of burn injury clearly effect prognosis, the target population is fairly easily defined by the trauma of burn injury. The aggregation of 60% of burn admissions at 128 US burn centers should improve ability to access this population for research.¹ Severely injured burn patients often require multiple surgeries or long inpatient stays that make them a captive audience for recruitment. In addition, burn centers provide very specialized treatment paradigms by consistent faculty standardizing care for the patient population seen in that center.

However, barriers to research participation may exist in the burn population that are not present in other critically ill populations. The shocking nature of burn trauma and the associated consequences beyond injury, like unstable housing, may make encounters about research unbearably overwhelming. Extreme pain may limit a burn patient’s ability to focus on a discussion of research risks and benefits. While we did not examine causality, we observed that SDMs and patients were more willing to hear about research if we waited to approach them approximately 48 hours after admission and when pain was adequately controlled. Mental health diagnoses, education and income level, and motivation for enrollment were not collected in the pre-enrollment period of our study, but these factors are also likely to be important in fully understanding barriers to participation in burn research and the ability to consent.

SDMs play a crucial role in studies of critically ill patients. Eighty-seven percent of patients surveyed after critical illness describe research consent provided by a SDM as acceptable or highly acceptable.⁴ SDM rationale for providing or declining consent in the ICU has been previously described.⁸ Frequent reasons that SDMs provide consent include 1) belief that the patient will benefit from participation; 2) a strong desire to help others; 3) wanting to advance medical knowledge; and 4) the belief that their loved one would want to participate. The most common SDM reasons for declining consent were 1) uncertainty about their loved one’s wishes regarding research participation and 2) the concern that the risk may be greater than the benefits for their loved one.

In our experience, we found that SDM often expressed a “need to discuss it [research] with other family members.” We were surprised that this seemed to apply to both the non-invasive study as well as the study with minimally invasive specimen collection. From our perspective, SDMs seemed to want to discuss with additional family members the more general concept of whether the patient would be willing to participate in any type of research rather than the specific patient data or samples that would be collected throughout the studies. Of the SDMs that gave consent, none of them consented to the observational study alone. If the SDMs provided consent, they consented to both the observational and minimally invasive studies. This suggests that the reason for declining consent did not stem from fear surrounding the discomfort or distress due to surveillance specimen collection but rather due to other reasons.

Physicians also likely play an important role in the decision to participate in burn research. Physicians have been described as both a barrier to study enrollment for failing to refer eligible patients to clinical trials as well as a motivator for participation in trials because patients trust their doctor’s recommendation.⁹ A study of physician vs. non-physician approach of SDMs in the ICU for consent found that 87.5% of SDMs felt that physician involvement was inconsequential.⁸ Instead, SDMs prioritized the personal attributes of the person approaching them over professional designation. Another study found that 85% of SDMs indicated that a medical update from an ICU physician prior to a discussion about research would be helpful and 53% wanted the treating intensivist involved in the research decision-making process.⁵ When designing our study, we chose not to involve the burn surgery ICU attending in recruitment given concern that their involvement could be considered coercive. However, these findings of Barrett et al. suggest that SDMs may not perceive ICU attending involvement in recruitment as such.

As our study progressed, the physician PI became increasingly involved in recruitment although no patients were officially consented by the PI. In our experience, SDMs and patients seemed more likely to return phone calls and to be willing to listen about research introduced by a physician than a research coordinator. Introduction of the research topic by the investigating physician could influence decision-making in several possible ways: 1) by making the research topic appear more important or more interesting; 2) by making research participation sound more appealing; or 3) by creating a feeling of desire to express gratitude to the care team. Some or all of these factors may have contributed the decreased time to enrollment noted in the late period of our study.

Finally, the Canadian Clinical Care Trials Group recently published the results of The Consent Study in 2013 describing challenges of implementing acute care research.¹⁰ The most notable conclusions from this study were that 1) a large gap existed between eligibility and the frequency of consent encounters in the ICU and 2) recruitment was highly susceptible to design and procedural inefficiencies and to personnel availability. These findings are very consistent with what we experienced our study. The patients who died prior to recruitment in the early period our study were likely due to missed or infeasible opportunities for recruitment or to the challenges of approaching SDMs of critically ill patients about research by phone and the limited availability of staff to meet with them in person.

As this is an observational study, we cannot conclude with certainty that all differences in time to enrollment were directly related to the interventional modifications. However, given that the process modifications were directed toward perceived weaknesses, the intervention likely did have an important effect. Three additional factors varied between the study periods that limit our ability to attribute the full effect to our intervention alone. First, fewer eligible patients were admitted to the burn ICU during the later study period, which may have led to fewer demands on the recruitment team during this time. This explanation is unlikely to be sufficient to explain all of our findings since multiple research staff members participated in the earlier recruitment period whereas a single research coordinator recruited in the later period. Second, a larger proportion of patients in the second period were able to consent for themselves, which may have made them more easily accessible to the recruitment team. However, since SDMs could be recruited by telephone, SDMs may have been as accessible for recruitment as hospitalized patients who could be occupied by procedures or therapy or be unwilling to discuss research due to pain. Lastly, the increased confidence and experience of the study team, including the research coordinator and PI, in the later period is likely to have contributed this apparent improvement.

Successful clinical trials involving burn patients are possible. Most of the evidence-based practices in burn care are from very successful single centers that have perfected the research enterprise with a defined patient population, standard treatment algorithms and a consistent surgical team.^11,12 Such centers have well-established basic science, animal, and translational science labs,^13,14 perform randomized-controlled trials to study treatment interventions,¹⁵ and spearhead multi-centered trials.^16,17 These centers have established research protocols describing the risks, benefits, alternatives, and evidence, and have the staff to provide information and resources for patients and families with questions or concerns. The experiences of these centers have educated them on the many pitfalls for which they are now prepared. We encourage these centers to describe their path to success and to advise young investigators how best to proceed in conducting clinical trials in burn.

In conclusion, modifications to our recruitment process appeared to successfully decrease time to enrollment of severely injured burn patients into non-interventional studies. Greater research coordinator experience, flexible availability of study personnel for consent approach, and broad time windows for study inclusion are important to limit low accrual. Formal studies, including surveys of burn patient attitudes and beliefs, will be important to better understand barriers to recruitment and motivation for study enrollment in this unique population.