Recruitment Into a Pediatric Continuous Glucose Monitoring RCT

Abstract

Background:

The purpose was to identify patient/family characteristics and recruitment process characteristics associated with the decision to participate in a 2-year continuous glucose monitoring (CGM) RCT for youth with type 1 diabetes and their families.

Method:

Study staff approached patients who were conditionally eligible according to medical record review or referred by a provider. We categorized families according to participation decision (agree vs decline) and timing of decision (day of approach vs later [“thinkers”]).

Results:

Over 18 months, we approached 456 eligible patients; 19% agreed on the day of approach, 10% agreed later, 42% declined on the day of approach, and 30% declined later. Agreers were younger (P = .002), had shorter diabetes duration (P = .0003), had a lower insulin dose (P = .02), checked blood glucose levels more often (P = .002), and were more likely to use pump therapy (P = .009) than decliners. Patients/families were more likely to agree in fall/winter (41%) than spring/summer (19%, P < .0001). Of decliners, 50% cited no interest in CGM as the reason for nonparticipation. Among thinkers, 49% of patients who made a decision within 2 weeks of being approached agreed; only 15% of thinkers who made a decision >2 weeks after being approached agreed to participate (P < .0001).

Conclusions:

Recruitment is a critical and often challenging phase of clinical trials. Recruitment to pediatric CGM studies may be especially challenging due to youths’ reluctance to use CGM. These data provide an opportunity to better understand and possibly optimize recruitment into future pediatric CGM studies and other studies of advanced diabetes technologies.

Participation in pediatric clinical studies often requires substantial time commitment for families already challenged with managing chronic conditions like type 1 diabetes in their children.¹ Patients and families must balance the potential benefits of study participation (eg, opportunities to try new technologies and/or treatments) with the time commitment and potential risks.^2-5

For youth with type 1 diabetes and their families, clinical studies provide opportunities to try new diabetes technologies, such as continuous glucose monitoring (CGM). Research has demonstrated improved glycemic control with consistent CGM use in patients with type 1 diabetes.⁶ However, it is challenging for youth with type 1 diabetes, especially adolescents, to maintain the level of CGM use associated with improvements in glycemic control.^7,8 Barriers to sustained CGM use include pain with sensor insertions, alarms, body concerns (eg, size of device), and technical issues.^9-11 Although advances in CGM technology aim to improve accuracy and ease of use and thus minimize such issues,^12,13 youth with type 1 diabetes remain reluctant to use CGM because of such concerns. Additional research is needed to find ways to overcome such barriers and achieve maximal benefit from CGM for pediatric patients.

The need for further research regarding pediatric use of CGM necessitates the investigation of ways to enhance recruitment of youth with type 1 diabetes and their families into clinical studies involving CGM. Type 1 diabetes offers both opportunities and challenges for recruitment to clinical studies. The quarterly frequency of routine type 1 diabetes clinic appointments provides an opportunity to schedule study visits on the same days as clinic appointments to minimize travel time for study visits. Parents may also be motivated by the potential health benefits for their child and by the opportunity to contribute to the field of diabetes research.³ On the other hand, type 1 diabetes involves a demanding daily regimen of blood glucose monitoring, insulin administration by injections or pump, and attention to dietary intake and physical activity, on top of the usual school and social activities of childhood and adolescence. Therefore, patients and families may be reluctant to commit the time and effort required for participation in a clinical study.¹⁴

Detailed recruitment data and reasons for nonparticipation have not frequently been reported for longitudinal pediatric CGM studies. Greater reporting may help lead to a more comprehensive understanding of the willingness of youth with type 1 diabetes to use CGM, specifically in the setting of clinical studies. An example of such reporting is the CGM Time Trial (Timing of Initiation of Continuous Glucose Monitoring in Established Pediatric Diabetes), in which youth aged 5-18 years, with type 1 diabetes for ≥1 year, were randomized to pump start with simultaneous CGM start or pump start with CGM start after 6 months.¹⁵ Of the 353 patients screened by investigators, approximately 12% were ineligible because they were not willing to use CGM. Data such as these are needed to better understand the challenges to implementing CGM in youth with type 1 diabetes.

Recruitment is a critical phase of any clinical study. Difficulties during recruitment have the potential to delay the start of a trial, reduce the power for outcome analyses, impact the generalizability of results to other samples, and increase costs. Our aim was to identify factors related to enrollment into a 2-year randomized controlled clinical trial (RCT) comparing standard CGM education to standard CGM education plus a family-focused psychoeducational intervention. Specifically, we sought to identify (1) patient/family characteristics and (2) recruitment process characteristics that were associated with families’ decisions regarding study participation.

Methods

We recruited youth with type 1 diabetes and their parents to participate in a 2-year RCT comparing CGM use with or without a family-focused psychoeducational intervention to overcome barriers to sustained CGM use.¹⁶ The Institutional Review Board approved the study protocol prior to the start of recruitment. Eligibility criteria for the RCT included age 8-17 years, type 1 diabetes duration ≥1 year, A1c 6.5-10.0% (48-86 mmol/mol), insulin dose ≥0.5 units/kg/day, blood glucose (BG) monitoring frequency ≥4 times/day, no consistent real-time CGM (RT-CGM) use in the last 6 months, and ongoing care at the clinic. A complete list of eligibility criteria is available at www.clinicaltrials.gov (NCT01472159).

Participation in the RCT required 12 study visits, each lasting approximately 1-1.5 hours, over 2 years; most visits could occur on the same days as routine clinic appointments, with 3 additional visits between clinic appointments during the first 6 months. Study visits involved routine CGM education with a study nurse, BG meter and CGM device downloads, parent-youth interviews (for diabetes management and CGM usage self-report), surveys every 6 months, chart reviews (completed by study staff after the visit), and psychoeducational intervention sessions for those randomized to the intervention group. The study provided CGM devices/supplies for the first year and study staff assisted families in obtaining insurance coverage for devices/supplies for the second year. Families received free parking for study visits and youth received $20/visit as compensation for their time and effort. Parents/youth who agreed to participate in the RCT provided informed consent/assent before beginning any study procedures. We obtained the primary reason for not participating from families who declined. Study staff coded each open-ended response into 1 of 7 categories: no interest, lack of time/not a good time, privacy concerns, too many extra visits, distance, CGM, and other.

Recruitment Methods

We used multiple methods of recruitment. The most common method was for study staff to approach families at regular clinic visits. We also used IRB-approved recruitment materials (printed flyers in clinic, study ad in printed/electronic pediatric newsletter, registration on www.clinicaltrials.gov) that included a phone number and email address for families to contact study staff about the study.

Each week, study staff reviewed the clinic schedule to look for patients in the study’s eligible age range. Study staff then reviewed these patients’ most recent clinical data and lab results in the electronic medical record and approached those who were eligible according to medical record review at their next clinic appointment to explain the study and confirm eligibility. A1c levels were assayed by DCCT-standardized immunoturbidimetric methodology using the Roche Cobas Integra assay (Roche Diagnostics, Indianapolis, IN). Study staff also approached families referred by clinic providers. Clinic providers referred families who expressed interest in the study during a clinic appointment or who the provider thought would benefit from CGM use and participation in the study. Study staff approached as many families as time/resources allowed each day. Due to staff turnover during the recruitment period, 5 different research assistants participated in recruiting patients/families.

Categorization of Recruitment Status

After each recruitment contact, we categorized the family’s recruitment status. Families who agreed to participate in the study on the day they were first approached by study staff were categorized as “immediate agree.” Families who declined to participate in the study on the day they were first approached by study staff were categorized as “immediate decline.” Families who did not decide whether or not to participate on the day they were first approached by study staff were categorized as “thinking.” After they agreed or declined, thinkers were categorized as “delayed agree” or “delayed decline,” respectively.

There are a few exceptions to these categorizations. Two families declined but then contacted the study team at a later date to express interest in the study. Four families declined but then expressed interest ≥6 months later after being referred by their clinic provider. For these 6 families, as there was no intervening contact from the study team per the standard operating procedures for families unable to decide at the first approach, only the second set of recruitment contacts are counted in analyses for this article. In addition, 10 families agreed but then declined at a later date; these families are categorized as “thinking” for the time that they had agreed.

Study staff sent additional information about the study to the thinkers within 1 day of the initial recruitment approach and then followed up with the family after 1 week. If the family had not made a decision about participation at that time, study staff followed up with the family after 1 additional week (2 weeks after the initial approach). If the family still had not made a decision at that time, study staff followed up with the family at successive clinic appointments until the family made a decision about study participation, the youth turned 18 (and was no longer eligible), or recruitment ended. Study staff also contacted families in response to any family-initiated communication. Follow-up contacts were made by phone or e-mail according to each family’s preference.

Statistical Analysis

Data analyses were performed with SAS software, Version 9.2 (SAS Institute, Inc, Cary, NC, USA). Unpaired t-tests and chi-square tests were used to compare groups. The nonparametric median test was used to compare number of follow-up contacts between delayed agreers and delayed decliners. Values are presented as mean ± SD unless otherwise indicated.

Results

Over 18 months, we screened 957 patient records for eligibility. Of these, we approached 456 eligible patients to reach our target enrollment of 130 (Figure 1). An additional 335 patients were deemed ineligible and 166 patients were not approached due to limited staff resources or cancelled appointments. Only 29% of the 456 eligible/approached patients agreed to participate. Of those approached, 86 (19%) agreed on the day they were first approached (immediate agree), 44 (10%) agreed on a later day (delayed agree), 190 (42%) declined on the day they were first approached (immediate decline), and 136 (30%) declined on a later day (delayed decline).

Figure 1.

Flow chart of recruitment.

Recruitment Characteristics

Those who agreed to participate, either when first approached or at a later time, were younger (t = −3.05, P = .002), had shorter diabetes duration (t = −3.64, P = .0003), had a lower insulin dose (t = −2.38, P = .02), and were checking BG levels more often (t = 3.17, P = .002) than those who declined (Table 1). In addition, a higher proportion of those who agreed were using pump therapy than those who declined, χ²(1) = 6.93, P = .009 (Table 1). There were no significant differences in sex or A1c between those who agreed and those who declined.

Table 1.

Characteristics of Eligible/Approached Patients.

	Immediate agree (n = 86)	Delayed agree (n = 44)	ALL AGREEa (n = 130)	Immediate decline (n = 190)	Delayed decline (n = 136)	ALL DECLINEb (n = 326)	P c
Age (years)	12.5 ± 2.8	12.9 ± 2.4	12.6 ± 2.7	13.4 ± 2.6	13.5 ± 2.6	13.5 ± 2.6	.002
Diabetes duration (years)	5.9 ± 3.5	6.0 ± 3.7	5.9 ± 3.5	7.5 ± 3.6	6.9 ± 3.2	7.2 ± 3.4	.0003
Sex (% female)	47	59	51	49	45	47	.53
Daily insulin dose (units/kg)	0.8 ± 0.2	0.9 ± 0.3	0.8 ± 0.2	0.9 ± 0.2	0.9 ± 0.2	0.9 ± 0.2	.02
Insulin regimen (% pump)	83	89	85	73	74	74	.009
BG monitoring (checks/day)	7.2 ± 2.3	6.9 ± 2.1	7.1 ± 2.2	6.1 ± 2.0	6.6 ± 2.3	6.3 ± 2.1	.002
A1c							.13
%	7.9 ± 0.8	7.9 ± 0.7	7.9 ± 0.8	8.1 ± 0.8	7.9 ± 0.8	8.0 ± 0.8
mmol/mol	63 ± 9	63 ± 8	63 ± 9	65 ± 9	63 ± 9	64 ± 9
Previous CGM use							.02
RT-CGM (%)	5	11	7	12	14	13
Masked CGM (%)	13	11	12	7	5	6
Both (%)	7	0	5	2	2	2
None (%)	76	77	76	79	79	79

Values are mean ± SD or percentage.

^aAll agree is the aggregate of immediate agree and delayed agree.

^bAll decline is the aggregate of immediate decline and delayed decline.

^cAll agree vs all decline.

Of the 456 eligible approached patients, 22% (n = 100) had used RT-CGM (n = 51), masked CGM (n = 36), or both (n = 13) previously. None of the eligible patients were currently using CGM consistently, as this was an exclusion criterion. Only 18% of patients who had used RT-CGM agreed to participate compared to 44% of patients who had used masked CGM, 46% of patients who had used both, and 28% of patients who had used neither, χ²(3) = 9.48, P = .02.

Families were significantly more likely to agree during the fall/winter months (October-March) than during the spring/summer months (April-September), χ²(1) = 22.91, P < .0001. The percentages of patients who agreed to participate in each season were 41% in October-December, 40% in January-March, 17% in April-June, and 20% in July-September (Figure 2).

Figure 2.

Percentage of eligible/approached patients who agreed versus declined according to month of decision. Black bars = agree; gray bars = decline. Patients were significantly more likely to agree in fall/winter months (October-March) than in spring/summer months (April-September) (P < .0001).

Reasons for Decline

Of the 326 patients who declined, 50% (n = 162) (or 36% of all 456 eligible approached patients) cited CGM as the primary reason for not participating (Table 2). Some of the CGM-specific reasons cited by these patients/families included: did not want another insertion site, did not want another device to carry, tried CGM in the past and did not like it, and wanted to wait until the technology is better. The other primary reasons for declining were lack of time/not a good time (13%), no interest (13%), distance (6%), too many extra visits (3%), and other (4%) (Table 2). The remaining 11% of those who declined were “passive decliners” who never made an explicit decision about participating in the study; study staff followed up with them according to standard recruitment procedures until they turned 18 (n = 3) or recruitment ended (n = 34).

Table 2.

Reasons for Decline

	n	%
CGM	162	50
Lack of time/not a good time	43	13
No interest	42	13
Distance	19	6
Too many extra visits	9	3
Other	14	4
Passive decline	37	11

Notably, there was no difference in previous CGM use between decliners citing CGM as the reason for study refusal (24%) versus those declining for other reasons (19%), χ²(1) = 1.12, P = .29. Younger youth were more likely to decline because of CGM and older youth were more likely to decline because of time, χ²(2) = 9.46, P = .009: 60% of decliners ages 8-14 years cited CGM compared to 46% of decliners ages 15-17 years and 11% of decliners ages 8-14 years cited lack of time/not a good time compared to 25% of decliners ages 15-17 years.

Thinkers

Figure 3 shows the distribution of time between date of first approach and date of decision for the 180 thinkers. Decision time was shorter for delayed agreers (median = 2 weeks, range = 1-49 weeks) than delayed decliners (median = 13.5 weeks, range = 1-71 weeks). About one-quarter of thinkers (n = 51, 28%) made a decision within 2 weeks of being approached, half of whom (n = 25) agreed to participate; of the remaining 129 (72% of the thinkers) who made a decision >2 weeks after being approached, only 15% (n = 19) eventually agreed to participate, χ²(1) = 23.27, P < .0001.

Figure 3.

Number of months between date of initial approach and date of decision for thinkers. Black bars = delayed agree; gray bars = delayed decline. Among thinkers, patients who made a decision within 2 weeks of initial approach were more likely to agree (P < .0001) and required fewer follow-up contacts (P < .0001) than patients who made a decision after 2 weeks.

Study staff made a median of 2 follow-up contacts (range = 1-6, mode = 1 and 2) to delayed agreers and a median of 3 follow-up contacts (range = 0-14, mode = 3) to delayed decliners, χ²(1) = 5.34, P = .02. This difference was explained by the longer decision time among delayed decliners than delayed agreers and the significant association between longer decision time and more contacts from study staff independent of participation decision. Only 27% of thinkers who made a decision within 2 weeks of being approached required 3 or more contacts from study staff compared to 61% of thinkers who made a decision after 2 weeks, χ²(1) = 16.71, P < .0001.

Discussion

Recruitment is a critical but often challenging phase of a clinical trial as researchers try to enroll a targeted number of participants within a limited period of time. The reluctance to participate that many researchers encounter when recruiting for pediatric studies may be heightened when recruiting for pediatric CGM studies. Despite the demonstrated glycemic benefits associated with sustained CGM use, the majority of youth with type 1 diabetes do not use CGM.¹⁷ In this study, less than one-third (29%) of eligible patients who we approached agreed to participate. We identified several factors associated with families’ decisions to participate or not participate in a 2-year clinical trial involving CGM.

Patients who agreed to participate were younger, had shorter diabetes duration, had lower daily insulin dose, were checking BG levels more often, and were more likely to be using pump therapy than patients who declined. Similarly, youth with type 1 or type 2 diabetes who agreed to participate in the epidemiologic SEARCH for Diabetes in Youth study were more likely to be younger and have shorter diabetes duration than those who did not participate.¹⁸ SEARCH study participants were also more likely to be of non-Hispanic white race than nonparticipants.¹⁸ Herbert et al similarly reported that youth with type 1 diabetes who enrolled in any of 3 behavioral intervention studies were more likely to be younger and of Caucasian race than youth with type 1 diabetes in the general clinic population.¹⁹ One limitation of our study is that we do not have race/ethnicity or socioeconomic status of nonparticipants and therefore could not compare race/ethnicity or socioeconomic status between those who agreed and those who declined to participate. When resources are limited, knowledge of patient characteristics associated with willingness to participate in CGM studies may allow study staff to focus their efforts on recruiting patients who are most likely to participate.

Half of those who declined cited CGM as their reason for not participating. Some of these patients indicated that they did not want another insertion site or another device to carry. Other patients said they had had unfavorable experiences with CGM in the past and did not want to use it again or wanted to wait for improvements in the technology. The high percentage of those who cited CGM as their reason for declining highlights the negative influence of patients’/families’ perceptions of CGM, often informed by previous exposure to earlier devices possessing inferior or suboptimal performance. It also highlights the challenges of enrolling pediatric patients and their families into CGM studies due to the recognized barriers to CGM use in youth with type 1 diabetes.^{9,10,13,20,21}

It is important that researchers implement procedures for following up with thinkers in a timely manner in the immediate days/weeks after the initial recruitment approach. In our study, thinkers received 3 contacts from study staff in the 2 weeks following their initial approach in-clinic. Patients/families who did not make a decision on the day they were approached were 5½ times more likely to agree if they made a decision within 2 weeks than if their decision time was longer, highlighting the importance of timely follow-up with thinkers. Some patients/families may be hesitant to agree to participate upon initial invitation; thus they want time to “think” about participation. Delays in the patient/family determination of participation lead to expenditures of staff time and resources to continue contacting these families regarding possible study enrollment. Overall, 66% of participants agreed to participate upon their initial contact, 19% within 2 weeks of the initial approach, and 15% up to 12 months later. In contrast, 58% of nonparticipants declined upon initial contact, 8% within 2 weeks, and 34% up to 17 months later. Thus, if there is no shortage of potentially eligible patients for study participation, as can be found at many large pediatric diabetes centers or in multicenter clinical trials, it is likely more cost-effective to eliminate further patient/family contact beyond the initial 2 weeks. Such patients/families can be provided with information to contact study staff if their desire to participate changes.

Eight percent of eligible/approached patients never made a decision before recruitment ended or before they turned 18 and were categorized as passive declines. Whittemore et al experienced an even higher rate of passive declines when recruiting for a multisite study involving 2 Internet-based psychoeducational programs for youth with type 1 diabetes, ages 11-14 years.²² Of the 518 patients that they approached, 22% declined when approached and 15% agreed but never completed the baseline assessment and were considered passive declines.

The rate of agreement to participate was twice as high in the fall/winter months as in the spring/summer months, suggesting a seasonal variation in families’ willingness to begin participating in a CGM study. When starting CGM, patients and families must invest time and energy into learning the technology and how to use the data it provides; they may be more willing to do this at some times of the year than at others. Some families declined because they did not want to start CGM during a particular sports season or at the same time that they were starting another technology, like an insulin pump. When patients/families decline study participation for temporal reasons like these, it may be beneficial to provide them with contact information regarding the study or for the research staff to ask the families if they would like to be approached again in the future (eg, 3-6 months later). Unfortunately, we did not follow this procedure in the current study. However, 6 patients initially declined but then contacted study staff or were referred by their clinic provider at a later time and 3 of them agreed to participate.

When a family is considering participation in a clinical study, it is important that the youth is involved in the decision at a level appropriate for his or her cognitive and emotional competence.^23,24 This is even more important when the study involves the youth’s use of an intensive management tool such as CGM. Indeed, the American Diabetes Association (ADA) and the International Society for Pediatric and Adolescent Diabetes (ISPAD) have stated the importance of individuals’ readiness and desire to use CGM. According to ADA clinical guidelines, “optimal CGM use requires an assessment of individual readiness for the technology.”²⁵ Furthermore, an ISPAD consensus statement recommended that youth “must have a personal interest in wearing [a CGM device] and not be a passive accomplice to a parental decision.”²⁶ After being approached, many parents said that they wanted more time to discuss the study with their child. In addition, many of the parents who declined participation specifically mentioned that their child did not want to wear the CGM device. These examples provide anecdotal illustrations of some youths’ involvement in the decision-making process about study participation. To better understand youth involvement in decisions regarding study participation, recruitment for future studies should include more systematic data collection regarding youth versus parent interest in study participation.

Understanding optimal recruitment strategies for pediatric technology studies will increase in importance as there will likely be a number of upcoming pivotal trials involving variations of automated insulin delivery systems over the next few years in which CGM plays an important role.²⁷ While we identified a number of factors associated with CGM study participation and a time efficient strategy to focus on patient enrollment, our report has some limitations. Recruitment experiences may be different in multicenter clinical trials and warrant separate investigation to maximize patient participation. There may be recruiter characteristics or factors related to recruitment methods that influence study uptake that were not considered; however, training of recruiters was consistent across time. Finally, while it is important to recruit pediatric patients/families efficiently for CGM studies, one must always approach patients with respect and provide them with an opportunity to decline participation in accordance with standards for the protection of human subjects.

Conclusions

In summary, only 29% of eligible patients who we approached agreed to participate; 50% of those who declined specifically said that they did not want to use CGM. Thus, for many youth with type 1 diabetes, the barriers to CGM use appear to outweigh the potential glycemic benefits, whether for research or clinical use. Families were more likely to enroll in this pediatric CGM study in fall/winter than spring/summer months. These data also highlight the importance of following up with families in the immediate days and up to 2 weeks after the initial recruitment contact due to the diminishing return of effort with longer time of indecision. Although many publications have described the general challenges associated with recruitment into clinical trials, few articles have described detailed recruitment data for pediatric CGM studies. The data provided by this observational assessment of the recruitment process for a 2-year CGM RCT may provide an opportunity to better understand and optimize recruitment into future pediatric clinical trials involving CGM and possibly other advanced diabetes technologies.