Closing the Gap: Results of the Multicenter Canadian Randomized Controlled Trial of Structured Transition in Young Adults With Type 1 Diabetes

Transitional care is the purposeful, planned movement of adolescents and young adults with chronic conditions from pediatric to adult care health care systems (1). Emergent adulthood is a challenging period of increasing maturity, independence, and personal identity (2). Many health care systems require that pediatric patients be transferred to adult care at ∼18 years of age, which is also a time of major changes in school, work status, and residence (3). Differences in child-centered and adult health systems contribute to the abrupt change (1) and dissatisfaction with care (4,5).

For young adults with type 1 diabetes, the additional stress associated with managing their disease (3,6) together with other competing demands can lead to a decline in metabolic control (7,8). Other possible factors that negatively impact control include insulin resistance associated with pubertal hormonal changes, reduced blood glucose monitoring, intentional insulin omission for weight management, risk-taking behavior, or mental health issues (9–11). As many as 30% of young adults with type 1 diabetes disengage from care during transition and 46% report difficulties with the transition process (12–14). Clinic attendance declines (15,16), and among those lost to follow-up, the HbA_1c is 1.5% higher than among those who remain in medical care (17). There is also evidence that challenges and behaviors experienced during transition track into adulthood (4,14), particularly for those patients with identified predictors of risk (18–20). Suboptimal glycemic control in young adults with type 1 diabetes increases the risk for hospitalization and diabetes-related complications (13,21) and increases the burden on health care systems.

Studies that have evaluated transitional care in young adults with type 1 diabetes have identified a need for more support during the transition (4,7). However, evidence for the types of supports that are effective in improving this important care gap is unclear (22). Moreover, although several interventions have been assessed (4,17,23,24), there have been no multicenter randomized trials that have tested the effect of transition strategies on follow-up visit adherence and glycemic control after transfer of care. We report results of a multicenter randomized trial that compared posttransition clinic attendance, HbA_1c levels, satisfaction with care, diabetes-related distress, and impact of diabetes on quality of life between young adults with type 1 diabetes receiving usual transition care or a structured transition program in which involvement of a transition coordinator (TC) over 18 months was a key component.

The study was a multicenter, randomized, parallel-group, controlled trial conducted in three pediatric (two tertiary and one secondary) centers together with their usual adult care referral centers in Ontario, Canada. The study was approved by each site’s local institutional review board, and informed consent was obtained from all study participants. The main inclusion criteria were patients with type 1 diabetes aged 17–20 years who had at least one clinic visit with their pediatric endocrinologist in the previous 12 months and who were scheduled to be transferred to adult diabetes care in the next 6 months. The full study protocol was published previously (25).

The transition program was designed to provide additional support during transition of care and was introduced 6 months prior to referral to adult diabetes care. The intervention was 18 months long and spanned six clinic visits (three pediatric and three adult). There were also two clinic visits in the 12 months of follow-up, so the total number of study visits was eight. Central to the program were TCs at each site who provided a link between pediatric and adult diabetes care. The TCs were Certified Diabetes Educators who provided transitional education and clinical support where appropriate. The role of the TC was to assist participants during the visits in the first 18 months, maintain contact with participants between the visits (by phone, text messages, or e-mail), facilitate support for insulin adjustments and sick day/hypoglycemia management during regular hours, send reminders and help reschedule appointments, and assess needs and facilitate referrals to other services (e.g., psychology, social work, nurse educator, or dietitian). The TCs also provided specific transition-related education and education materials and at the last pediatric visit a bio sketch of the adult endocrinologist to whom the participant had been referred as well as written instructions and maps to navigate adult diabetes centers.

Participants were expected to be seen in three pediatric diabetes clinic visits over 6 months followed by three adult diabetes clinic visits over 12 months. The TC was to attend all visits. Participants randomized to the standard diabetes care group followed the same clinic visit schedule with the only difference being the exclusion of additional support by the TC. Participants in both groups received diabetes care as per Diabetes Canada clinical practice guidelines (26).

At each of the eight visits (six visits during intervention and two during follow-up), data collected for both groups included clinic attendance; diabetes-related data including participant weight, height, BMI, blood pressure, insulin dosage, and method of delivery; and any interim diabetes-related emergency room, hospital visits, and adverse events. Also recorded was the frequency of complication screening (retinal, monofilament, lipid profile testing, and microalbumin-to-creatinine ratio). A local HbA_1c was done at visits one, two, three, four, and five (capillary and/or venous sample by the center’s usual method of testing). Central laboratory (DynaCare, Brantford, Ontario, Canada) HbA_1c measurements were done at visits one, six, seven, and eight using Diabetes Control and Complications Trial (DCCT)–validated assay. The complete details of visit assessments are in the study protocol, which was published previously (25).

Participants were randomly assigned in a 1:1 ratio to the transition program or standard care. The randomization schedule was computer generated in variable blocks stratified by HbA_1c (<8.5% [69 mmol/mol] or ≥8.5% [69 mmol/mol]) and site and held centrally at the data coordinating center. Because of the nature of the intervention, we could not blind participants and members of the diabetes treatment team to group allocation. The outcome assessors and data analysis personnel were blinded to the group assignment. There was no contact between the TC and the control group to minimize cross contamination.

We assessed outcomes during the 18 months of the intervention period and 12 months after completion of the intervention (follow-up period). The primary outcome was the proportion of participants who failed to attend at least one adult diabetes clinic visit during the 12-month follow-up period. Prespecified secondary outcomes were the frequency of HbA_1c testing, mean HbA_1c level, frequency of complication screening (nephropathy, retinopathy, and peripheral neuropathy), diabetes-related emergency room visits and hospitalizations for diabetic ketoacidosis (DKA) and hypoglycemia, patient satisfaction with the transition process, and diabetes distress and impact of diabetes on quality of life. The mean HbA_1c was calculated separately for the central and local measurements collected during the designated periods. Patient satisfaction with diabetes care was assessed using the Client Satisfaction Questionnaire (CSQ), diabetes distress using the Diabetes Distress Scale (DDS), and the impact of diabetes on quality of life with the Diabetes Quality of Life (DQL). These were done at baseline, completion of the intervention, and completion of follow-up. A transition intervention evaluation questionnaire was administered to those in the transition program at completion of follow-up. This 25-item survey evaluated the overall transition experience as well as perceived personal support from the TC, educational elements, and structural/navigation support.

The study sample size determination was based on the projected proportion of participants in each group who would fail to attend at least one diabetes clinic visit during the 12 months after the intervention period. We reviewed the records of 156 young adults who were transferred from the Children’s Hospital, London Health Sciences Centre, to adult care at St. Joseph’s Health Care London between January 2005 and December 2008 and found a rate of nonattendance at the adult diabetes clinic between 2 and 5 years after transfer of 28%. To detect an absolute difference of 16% (28% nonattendance rate in the control group compared with 12% in the intervention group), 188 participants (94 per group) were required to provide 80% power at the 0.05 level of significance. Because loss to clinical follow-up was the study’s primary outcome, we did not increase the sample size to allow for loss to follow-up.

The primary outcome was the proportion of participants who failed to attend at least one outpatient adult diabetes clinic visit during the 12 months after the intervention period. Between-group comparisons, adjusting for center and baseline HbA_1c (<8.5% [69 mmol/mol] or ≥8.5% [69 mmol/mol]), used the Cochran-Mantel-Haenszel score test. The primary analysis was based on the comparison of participants in the two treatment groups who attended zero, one, and two clinical visits in the 12 months after the 18-month intervention period.

To control for covariates of interest, the proportional odds model was adopted. Exploratory analysis was performed to assess possible interactions between the intervention and control groups and baseline variables (such as center effect, age of transfer, sex, HbA_1c level, distance to diabetes center, postsecondary education institution attendance and location, and clinic attendance rate/year prior to transfer) to examine if the effect of the intervention (transition program) on the rate of nonattendance was influenced by one of the prognostic factors (e.g., HbA_1c level), taking into account the baseline covariates. Questionnaires were analyzed using ANCOVA adjusting for site baseline HbA_1c group (<8.5% [69 mmol/mol] or ≥8.5% [69 mmol/mol]) and the baseline value of the end point. Difference in change from baseline scores was calculated between the groups based on ANCOVA adjusted for the same variables as above.

Four mixed logistic regression models were used to analyze the effect of the intervention adjusted for the study period and baseline HbA_1c level. The minimal Akaike information criterion was used as the indicator of the best fitted model. The statistical analysis was performed using SAS (version 9.4) and R (version 3.4.4). P values ≤0.05 were considered statistically significant.

Between 17 April 2012 and 24 July 2014, 466 participants in the three pediatric centers were assessed for eligibility and 205 participants were randomized to the structured transition program (n = 104) or to standard care (n = 101) groups. Baseline characteristics were similar between groups (Table 1). The mean number of pediatric diabetes clinic visits in the 12 months preceding the enrollment was 3.2 (SD 0.9) in each group. The number of clinic visits attended in the year prior to study entry did not have an effect on clinic attendance during the 12 months after the intervention completion (odds ratio 1.29 [95% CI 0.86, 1.94], P = 0.214) (Supplementary Table 3).

The following results are presented first for the intervention period (18 months) and then for the follow-up period (12 months) during which the primary outcomes were assessed. During the intervention period, there were more visits in the transition program than in standard care. The mean number of visits over 18 months was 4.1 (SD 1.1) in the transition program versus 3.6 (SD 1.2) in standard care (P = 0.002) (Table 2). There were 51 (49%) participants in the transition program and 26 (26%) participants in standard care who attended all six visits. There was no difference in the mean change in HbA_1c from baseline to 18 months between the two groups (adjusted difference −0.04% [0.40 mmol/mol] [95% CI −0.49, 0.40], P = 0.848) (Table 3).

There were more participants in the transition program than standard care who had at least one emergency room visit for a diabetes-related problem other than DKA and/or hypoglycemia during the intervention period (9 [9%] vs. 2 [2%], P = 0.002). There was no difference in hospitalizations or emergency room visits for DKA and/or hypoglycemia between the groups (data not shown). More participants had monofilament foot testing in the transition program than standard care during the initial 18 months (71.2% vs. 57.4%, P = 0.036). During the 18 months of intervention, there were on average 17.6 (SD 7.3) indirect contacts with the TC. In addition, the TC attended almost all six visits in person (mean number of in-person visits 5.0 [SD 1.4]). The majority of contacts occurred by text messages (7.4 [SD 8.1]), whereas e-mail and phone contacts were less frequent (3.2 [SD 3.9] vs. 1.7 [SD 2.8]).

CSQ and DDS questionnaires were completed by 71 (68.3%) participants in the transition program and by 57 (56.4%) participants in standard care. Participants who had access to the TC showed improved satisfaction with care (CSQ mean score 29.0 [SD 2.7] vs. 27.9 [SD 3.4], P = 0.032), less diabetes-related distress (mean score 1.95 [SD 0.8] vs. 2.18 [SD 0.8], P = 0.049), and less emotional burden of diabetes (mean score 2.3 [SD 1.1] vs. 2.7 [SD 1.2], P = 0.027) compared with their baseline scores. The change in DQL scores was similar between the groups (Table 4).

In the 12-month follow-up after the intervention period, 21 (20%) participants in the transition program and 20 (20%) in standard care were lost to follow-up. The mean number of visits, 1.3 (SD 0.8) in the transition program and 1.3 (SD 0.8) in standard care, was the same during the follow-up period (P = 0.846). Fifty-one (49%) participants in the transition program and 47 (47%) in standard care attended both follow-up visits (Table 2). There was no difference in the frequency of diabetes-related testing (retinal exam, albumin-to-creatinine ratio, lipid profile, and monofilament testing) in the follow-up period (Supplementary Table 1) or diabetes-related hospitalizations and/or emergency room visits. Questionnaires were completed by 70 (67%) participants in the transition program and by 68 (67%) in standard care. At completion of the study, there was no between-group difference in CSQ, DDS, DQL, or emotional burden of diabetes scores compared with baseline (CSQ score change difference −1.21 [95% CI −2.49, 0.07], P = 0.064; DDS scores change difference 0.06 [95% CI −0.20, 0.32], P = 0.642; emotional burden of diabetes scores change difference 0.05 [95% CI −0.28, 0.39], P = 0.756).

Post hoc analysis using a linear regression model showed that the odds of attending a clinical visit in the follow-up period was 1.3 times higher (95% CI 1.00, 1.79) in the intervention group (Supplementary Table 2). HbA_1c measurements were available for 73 (70%) participants in the transition program and 71 (70%) participants in standard care. Mean change in HbA_1c did not differ between the groups (adjusted difference 0.37% [95% CI −0.04, 0.78], P = 0.073) (Table 3).

The exploratory analysis showed that the mean HbA_1c at each visit in the control group was significantly higher than the mean HbA_1c at the same time point in the intervention group for the 30 months of study duration (mean difference 0.31% [95% CI 0.08, 0.54], P = 0.013) (Supplementary Fig. 2).

In the logistic regression model, we did not identify any baseline characteristics associated with a risk of disengagement with specialist care. To analyze the effect of the intervention on the clinic visits, the logistic mixed model with random intercept was applied (Supplementary Table 2). According to the prespecified criteria, model 3 was chosen as the “best.” It demonstrated that the odds of clinic attendance in the transition program was 1.34 (95% CI 1.005, 1.788, P value = 0.0457) higher than in standard care. The effect of the follow-up period had no effect on clinic attendance, with an odds ratio of 0.808 (95% CI 0.647, 1.009, P value = 0.06). The baseline HbA_1c level had no significant effect on clinic attendance.

In this multicenter randomized controlled trial (RCT), clinic attendance was improved in young adults with type 1 diabetes who received an 18-month structured transition program with a TC. They also reported greater satisfaction with care and experienced less diabetes-related distress than those in standard care. The mean change in HbA_1c during the intervention showed a trend toward improvement but did not reach statistical significance. The young adults who had been in the structured transition program and had improved their clinic attendance during this period did not attend more clinic visits during the year after the intervention. Secondary outcomes that had improved during the intervention period (satisfaction with care and diabetes-related distress) were no longer different in the follow-up period. Diabetes complication screening, DKA, and severe hypoglycemia episodes occurred at the same frequency in both groups during the 12-month follow-up period.

Although the number of clinic visits increased during the intervention, this was not associated with an improvement in glycemic control. Interestingly, HbA_1c did not increase in the intervention group, which was a trend seen in the control group. In the exploratory analysis, the mean of the difference in HbA_1c between the groups during the follow-up period was statistically significant, with the intervention group HbA_1c continuing to improve.

There were significantly more diabetes-related emergency room visits in the intervention group during the study intervention. We think that frequent contact with the TC may have resulted in higher awareness of the risk of acute diabetes-related complications and more frequent emergency room visits. It is reassuring that the intervention did not result in a change in the rate of serious adverse events (DKA/hypoglycemia).

We have shown that there are benefits of additional support during the transition period in young adults with type 1 diabetes. These benefits included improved clinic attendance, satisfaction with care, and less diabetes-related distress but did not translate to better glycemic control. We speculate that longer follow-up and/or a larger sample size would support the hypothesis that routine clinic visits result in better glycemic control and improved long-term diabetes-related outcomes.

To our knowledge, this is the first report of a multicenter RCT evaluating a transition intervention lasting for 18 months and extending to a 12-month follow-up. Strengths of our study included its robust design, pragmatic intervention, use of innovative, age-appropriate means of communication, and long follow-up duration spanning pediatric and adult care. To achieve an optimal study design, we built on various components of previously published transition interventions shown to improve health outcomes in observational studies (27,28) and aimed to maintain “real-world” conditions, with the only difference between the usual care and intervention groups being the addition of the structured transition program and TC. We successfully recruited and engaged this highly mobile and often challenging population throughout the transition period. The inclusion of two tertiary and a secondary clinical practice following young adults with type 1 diabetes allows for greater generalizability of our findings. The inclusion of electronic means of communication with the TC made our intervention more relevant and acceptable to young adults (29). Finally our intervention spanned pediatric and adult care and was longer than most reported transition interventions, which have been of 1 year duration or less.

Limitations of our study include lack of blinding and sample size, which possibly was not large enough to detect some differences. Blinding of participants and investigators was not possible; however, the group allocation was blinded, the treating physicians were not directly involved in the delivery of the transition-based interventions, and the TC had no contact with the control group participants.

Disengagement from care in the 12 months of follow-up was lower than anticipated and may have affected the ability to detect a difference between the groups at completion of the trial. We have shown that compared with the historic cohort in our center, rates of losses to clinical follow-up decreased in both groups. This can be explained by participation in a clinical trial and selection of a more motivated population that may perform better in a clinical trial setting.

A Cochrane Review in 2016 of registered controlled trials on transition care included adolescents with any chronic condition and any type of intervention and concluded that the limited number of evaluative studies with few randomized trials limits the strength of evidence (23,30). To our knowledge, there has been only one other RCT published since then, the Australian Transition to Adult Care in type 1 Diabetes (TrACeD) study (31), in which young adults with type 1 diabetes were assigned to an appointment manager group or to a control group. This study was smaller than ours, included a single center, and was not associated with increased clinic attendance in the first 12 months of the intervention but had a positive effect in the 2nd year (months 12–24). This is unlike our study, in which the improvement occurred in the first 18 months while the intervention was delivered but did not persist during the follow-up period when the extra support to young adults ended. Unlike the frequent TC/participant contact (an average of 17 contacts per participant over 18 months) in our study, in the TrACeD study, >50% of participants had infrequent contacts (defined as no contact over 12 months) with the appointment manager. Participant withdrawal rate was also much higher, with up to 38% of data missing for 2nd year outcome analysis. The TrACeD study was not powered to detect the difference in the follow-up period (12–24 months), and hence it is difficult to compare those results with our study. The number of participants in our trial exceeded all previous RCTs, including the TrACeD study. We are currently among very few reported RCTs to have studied the transition from both pediatric and adult perspectives. The only other RCT was an Australian pilot study in which young adults with type 1 diabetes were followed after discharge from pediatric care. The intervention included four contacts via phone from a TC over a 12-month period, together with educational and support materials (32). The sample size required for statistical power was determined at 30 individuals per group; however, results could only be reported on 14 in the control group and 12 in the intervention group due to challenges with recruitment and data collection. Hence, this study was also underpowered and in addition had a significant difference in HbA_1c between groups at baseline with minimal duration and frequency of phone contacts (32).

The important finding in our study was the significant improvement in clinic attendance in the intervention group during the initial 18 months when the transition program was delivered. Our study did not demonstrate a statistically significant improvement in the primary outcome, i.e., no difference in loss to clinic follow-up in the 12 months after the transition intervention. However, we have shown that the intervention can be effective. Our key finding speaks to the importance of ongoing transition support throughout the period of young adulthood. We also did not demonstrate an improvement in glycemic control in our study population. However, the mean HbA_1c in the intervention group did not increase over the study period (30 months) and may have shown some decreasing trend. Of note, our study population started with a mean baseline HbA_1c of 8.6% (70 mmol/mol) in the control group, compared with 8.9% (74 mmol/mol) in the same age cohort from the Type 1 Diabetes Exchange program (33). In our trial, disengagement rates were lower in both groups during the 12-month follow-up, which speaks to possible limitations of the clinical trial (recruitment of a highly motivated patient population). One of the most important results that should be emphasized is the improved satisfaction with care, decreased diabetes-related distress, and decreased emotional burden of illness. Any transition program that aims to change behavior should consider the emotional burdens of the disease in this population and aim to reduce barriers to access to care. We acknowledge that administering the transition program is time consuming, and inclusion of the TC (a Certified Diabetes Educator) is costly and may be a limiting factor to widespread implementation of this intervention. On the other hand, our intervention is easily translatable, and we ensured that insulin adjustments and medical care remained unchanged from usual care in our centers. Hence, the role of a TC could be modified and does not necessarily need to involve a diabetes educator or a highly specialized health care professional.

In summary, our study is the first to address the issue of transition support in a robust, pragmatic RCT. Young adults have a range of different needs related to their diabetes care and require flexibility when interacting with the health care system. We plan to build on our findings and develop technology-based solutions such as a “Virtual TC,” which could be incorporated into more prolonged transitional care. Although the need for transition intervention has been evident for many years, there has been a paucity of high-quality evidence on which to base transition recommendations and practice. We anticipate that the results of this appropriately powered RCT will help to inform a more complete and prolonged solution for transition. Our study provides support that a structured transition program can alleviate some of the emotional burden of diabetes and help young adults maintain regular contact with their diabetes clinics and health care providers. In our study, the benefits of the intervention were demonstrated for the duration of the intervention. Our hope is that with a longer duration, positive effects of the intervention may persist and facilitate individual development of the necessary skills to bridge the care gap. Our study provides support that a structured transition program beginning in pediatric care can contribute to bridging the care gap. Sustaining these improvements requires sensitization of adult health care providers to invest in prolonged support for young adults beyond the first 2 years posttransition, which will continue to provide benefits throughout their life span.

Acknowledgments. The authors acknowledge Keira Evans (St. Joseph’s Health Care London), Pamela Tyner (Trillium Health Partners), and Krystal Merrells (Children’s Hospital of Eastern Ontario) for development, help, and immense dedication to the role of TC. The authors thank nursing and administrative staff at St. Joseph’s Health Care London; Children’s Hospital, London Health Sciences Centre; Children’s Hospital of Eastern Ontario; The Ottawa Hospital; and Mississauga Trillium Health Center and Credit Valley Hospital, Ontario, Canada. The authors thank the Robarts Clinical Trial Institute staff (D. King, M. Chow, T. Alnowaiser, J. DePace, L. Sttit, and G.Y. Zou) for their support throughout the trial and Dr. B. Goldstein (posthumous), N. Concepcion, and O. Lou at JDRF Canada for their advice and support. Most of all, the authors thank all the young adults who gave up their time to participate in this study.

Funding. The trial was funded by JDRF (CCTN 1102) and grants under the JDRF Canadian Clinical Trials Network, a public-private partnership including JDRF and FedDev Ontario. T.S., T.R., E.G., P.G., I.H., M.L., J.Mal., J.Mah., D.M., A.P., A.S., R.S., A.U., and C.C. report grants from JDRF during the conduct of the study. M.L. reports grants from Western University during the conduct of the study.

Duality of Interest. T.S. reports personal fees from Novo Nordisk, personal fees from Sanofi Canada, personal fees from Bristol-Myers Squibb, personal fees from AstraZeneca, grants from Medtronic Canada, and grants from Lexicon Pharmaceuticals, Inc., outside the submitted work. I.H. reports grants and personal fees from AstraZeneca/Bristol-Myers Squibb; grants, personal fees, and other from Eli Lilly and Company; grants and personal fees from GlaxoSmithKline; personal fees from Roche Pharma; personal fees from Insulet Corp.; grants and personal fees from Janssen Pharmaceuticals; grants from Lexicon Pharmaceuticals, Inc.; grants from Medtronic; grants, personal fees, and other from Sanofi; personal fees from Takeda; grants, personal fees, and other from Merck; and grants, personal fees, and other from Novo Nordisk, outside the submitted work. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. T.S., E.G., M.L.L., J.Mal., A.P., A.S., and C.C. supervised the electronic case report forms, data collections, data quality control, and programming for the statistical analysis. T.S., E.G., P.G., I.H., M.L.L., J.Mal., J.Mah., D.M., R.S., and C.C. participated in subject recruitment. T.S., T.R., E.G., M.L.L., J.Mal., J.Mah., A.U., and C.C. assisted with data interpretation, further analyses of participant data, and manuscript revisions. T.S., T.R., and C.C. wrote the first draft of the manuscript. T.S., T.R., E.G., I.H., M.L.L., J.Mal., J.Mah., A.P., A.S., A.U., and C.C. contributed to the data interpretation and writing. T.S. and C.C. designed the study. I.H., M.L.L., J.Mal., and J.Mah. critically reviewed the design. T.S. and J.Mah. designed the statistical analysis plan. T.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. An oral abstract of this study was presented at the 77th Scientific Sessions of the American Diabetes Association, San Diego, CA, 9–13 June 2017, and this study was an invited presentation at the 20th Annual Diabetes Canada/Canadian Society of Endocrinology and Metabolism Professional Conference and Annual Meeting, Edmonton, Alberta, Canada, 1–4 November 2017.