Type 1 diabetes is a chronic disease affecting ∼3 million Americans (1). The disease is characterized by autoimmune destruction of insulin-producing cells; affected individuals depend on insulin therapy for survival. Although type 1 diabetes is the third most common chronic disease in childhood, 85% of individuals with type 1 diabetes are adults (2).
Optimal management of type 1 diabetes comes at the expense of heavy burden for those with the disease. Although major advances have been made both in the pharmacology of insulin analogs and technological tools to assist in glucose management, only 21% of individuals with type 1 diabetes are able to meet their glycemic targets (3). These struggles are more apparent during adolescence and young adulthood, a developmental period during which adherence levels are low and glycemic control is poorest. In fact, only 17% of youth achieve glycemic targets (3).
Type 1 diabetes increases morbidity and mortality from microvascular and macrovascular disease. The Diabetes Control and Complications Trial and its Epidemiology of Diabetes Interventions and Complications follow-up study examined the effect of intensive blood glucose control on the risk of future complications in patients >13 years of age with type 1 diabetes. These studies demonstrated that intensive blood glucose control profoundly reduces the risk of future microvascular complications (4–6).
Despite the many technological advances during the past several years, A1C has worsened rather than improving in the adolescent years, and the period from late adolescence through early adulthood remains a challenging time for youth with type 1 diabetes. The unrelenting day-to-day management of type 1 diabetes becomes particularly difficult (7) at this time of many transitions, as these emerging adults make important decisions about their education, career, and health care. Some move away from their home support to attend college, whereas others face the demands of work with or without college-related stress. Each scenario makes diabetes management challenging. Also, the transition from pediatric to adult health care providers can be a stressful process for patients and families.
High self-efficacy has been shown to improve the likelihood of achieving target glycemic control in adolescents and adults with diabetes (8,9). The Global TEENs Study showed that the diabetes health-related quality of life (D-HRQOL) demonstrated a strong association with glycemic control, with lower A1C correlating with higher D-HRQOL (10). In adults with type 1 diabetes, a diagnosis of depression increases the risk of severe hypoglycemic and hyperglycemic symptoms, and dysglycemia increases the risk of depression (11). Other factors that have been shown to affect glycemic control in adolescents include school, living situation, balancing of competing demands, and fitting into social circles (12). These observations highlight domains that may impact diabetes control in emerging adults with type 1 diabetes.
Whereas previous studies examined the effect of these factors independently in relation to glycemic control, we added participants’ living and work situations to gain a broader understanding of the interplay among multiple factors that have the potential to influence diabetes control. In this cross-sectional study, we quantified diabetes self-efficacy, QOL, and depression while exploring future educational and vocational plans for patients in the transition age-group.
Our goal was to evaluate self-efficacy, QOL, and depressive symptoms in adolescents and young adults in relation to diabetes control. Although technological advances have not translated into improved metabolic control, it is possible that they may create the premises for changes in the domains listed above. We hypothesized that 1) Individuals with higher self-efficacy and QOL and few or no depressive symptoms have better glycemic control and 2) living situation/parental support and work circumstances influence glycemic control independent of the reported self-efficacy assessment.
Research Design and Methods
Study Population
Individuals with type 1 diabetes who met inclusion criteria were approached sequentially during their routine follow-up visit at the Diabetes Center at Women and Children’s Hospital of Buffalo (currently Oishei Children’s Hospital) over a period of 8 months (June 2016 to February 2017). We included individuals 16–21 years of age who managed their diabetes with multiple daily injections (MDI) or continuous subcutaneous insulin infusion (CSII). Participants’ use of continuous glucose monitoring (CGM) systems was not recorded, although at that time >35% of patients in our clinic were CGM users. We categorized study participants into two groups: 1) teens aged 16–17 years who were in high school (11th or 12th grade) and 2) young adults aged 18–21 years who were high school graduates (up to 2 years post–high school graduation). The two groups were separated based on high school graduation because this time period is beset by significant life changes and challenges as patients move from their homes to attend college and/or work. We excluded participants with intellectual impairment that would preclude answering survey questions, non–English speaking participants, and those residing in a residential long-term care or mental health facility. There were no specific exclusion criteria regarding diabetes duration, A1C at diagnosis, or duration since diagnosis of type 1 diabetes.
This study was approved by the University at Buffalo institutional review board and was conducted in accordance with the Helsinki Declaration. Participants >18 years of age provided informed consent, whereas younger participants provided assent and their parent/guardian provided parental permission. Participants received $10 as compensation for their time to complete the surveys that were not part of standard of care.
Participant Parameters
Date of birth, date of type 1 diabetes diagnosis, and insulin management regimen were obtained from the electronic medical record. For each participant, point-of-care A1C was measured on the day the surveys were completed using a calibrated Alere Afinion Analyzer, a method certified by the National Glycohemoglobin Standardization Program for point-of-care A1C testing (13).
Study Surveys
Participants completed a brief questionnaire on paper that gathered information on current living situation, composition of household, and current and future educational and/or vocational plans. They were also queried about their current (for high school graduates) or future (for pre–high school graduates) plans regarding education, work, and living environment. These questions are regularly asked at follow-up visits as part of standard care.
Additionally, all participants completed three validated paper questionnaires: a depression screening (the Patient Health Questionnaire Modified for Teens (PHQ-9 Modified), the Stanford Self-Efficacy for Diabetes Scale (DSES) and a diabetes QOL measure (the Diabetes Quality of Life Brief Clinical Inventory (DQOL-BCI).
The PHQ-9 Modified is a depression screen that is routinely used in the adolescent population, including in those with chronic diseases such as type 1 diabetes. The PHQ-9 Modified (Cronbach α >0.8) (14) includes nine Likert scale questions to which participants answer from 0 = not at all to 3 = nearly every day. Additionally, there are questions that ask how a participant has been feeling over the past year as well as probe thoughts of suicidality in the past month or ever in life. Participants answered all the questions on the PHQ-9 Modified, including those assessing suicidality. Individuals with scores corresponding to the moderate or severe depressive symptom range (score >10) were referred to a clinical psychologist and/or social worker for psychological counseling.
The DSES is a reliable (α = 0.83), eight-question survey that asks participants to rate their confidence in performing activities related to diabetes care such as mealtimes, exercise, and hypoglycemia management (15). Responses range from 1 (not confident at all) to 10 (totally confident) that the individual could carry out a type 1 diabetes self-management activity. For this tool, higher scores indicate higher perceived self-efficacy. Lorig et al. (16) reported a mean score of 6.87 ± 1.76 in 186 participants.
The DQOL-BCI is a 15-item questionnaire that asks participants to rate such domains as satisfaction with their current diabetes treatment, the amount of time it takes to manage their diabetes, and how well they adhere to diabetes self-management activities if they are socializing with people who are not aware of their diabetes. Participants were advised that they could skip any survey question they did not feel comfortable answering. Although this survey has been validated and determined to be reliable (α = 0.85) in adults and not in adolescents, the questionnaire includes key discussion items that are part of standard care for youth with type 1 diabetes. Responses range from “very satisfied” to “very dissatisfied” for eight questions and from “never” to “all the time” for seven questions (17). Because the DQOL-BCI does not have a standardized scoring system, we adopted the scoring system of the Pediatric Quality of Life Inventory (18) to measure diabetes-specific QOL. This process entailed first scoring the answers based on a five-point Likert scale ranging from 1 = best DQOL (very satisfied) to 5 = worst DQOL (very dissatisfied), followed by a transformation to a scale ranging from 0 to 100, with 0 indicating lowest DQOL and 100 indicating highest DQOL. Higher scores indicated higher diabetes-related QOL.
Data Analysis
Data were expressed as mean ± SD and percentages. Mann Whitney U tests were used to assess differences between participants who were still in high school (HS) versus those who had graduated high school within the past 2 years (HSG). Spearman’s correlations were used to compute associations between DSES, DQOL, PHQ9, A1C, and time since diagnosis. Associations between DSES and A1C, as well as DQOL and A1C, were further tested using general linear models. Additionally, the estimated marginal means of A1C, adjusted for DSES, were computed after the addition of living arrangements (participants living with parents vs. separately) and vocational/work environment (participants not working vs. working part time or full time). The prevalence of mild, moderate, or severe depression among adolescents and young adults was calculated from the PHQ-9 Modified survey. Associations between PHQ-9 Modified score and A1C were tested in our participants. IBM SPSS Statistics v. 26 software was used for data analysis.
Results
Participant Characteristics
A total of 159 patients followed by a multidisciplinary team at the Oishei Children’s Hospital Diabetes Center were sequentially approached by research personnel, and 125 (78.6%) enrolled in the study. There was an even distribution of females and males. Eighty-six percent of participants identified as non-Hispanic White, whereas 14% identified as African American, Hispanic/Latino, American Indian, or Asian/Pacific Islander. The mean age was 17.7 ± 1.4 years, and time since type 1 diabetes diagnosis was 7.2 ± 4.3 years. Fifty of the 125 participants (40%) were HSG, and 49.7% of participants used CSII.
Glycemic Control and Survey Results
At the time data were collected, the American Diabetes Association (ADA) recommended goal for A1C was <7.5% for children (<19 years) and <7.0% for adults. Only 10.4% (13/125) met the ADA-recommended A1C goals. All participants received their diabetes care from a multidisciplinary team led by pediatric endocrinologists at a tertiary center.
Results of the questionnaires of the entire cohort, as well as for the HS and HSG groups separately, are summarized in Table 1. All 125 participants completed the three validated surveys, and their responses were included in data analysis. A total of 120 participants answered questions regarding living arrangements and vocational plans; three participants in the HS group and two in the HSG group did not complete the entire questionnaire. The five questionnaires that were incomplete were excluded from data analysis.
. | HS (n = 75) . | HSG (n = 50) . | Total (n = 125) . |
---|---|---|---|
DSES | 8.3 ± 1.4 | 8.2 ± 1.1 | 8.2 ± 1.3 |
DQOL-BCI | 74.3 ± 13.8 | 74.9 ± 11.1 | 74.3 ± 12.9 |
PHQ-9 Modified | 2.7 ± 4.2 | 3.0 ± 4.1 | 2.9 ± 4.1 |
A1C, % | 9.3 ± 1.8 | 9.1 ± 2.0 | 9.2 ± 1.9 |
(n = 72) | (n = 48) | (n = 120) | |
Living arrangements, %a | |||
Living with parents | 33 | 43 | 41 |
Not living with parents | 67b | 57 | 58 |
Vocational plans, %a | |||
Attending college | 83 | 77 | 81 |
Not attending college | 17c | 23 | 19 |
Attending college + workingd | 97 | 81 | 91 |
. | HS (n = 75) . | HSG (n = 50) . | Total (n = 125) . |
---|---|---|---|
DSES | 8.3 ± 1.4 | 8.2 ± 1.1 | 8.2 ± 1.3 |
DQOL-BCI | 74.3 ± 13.8 | 74.9 ± 11.1 | 74.3 ± 12.9 |
PHQ-9 Modified | 2.7 ± 4.2 | 3.0 ± 4.1 | 2.9 ± 4.1 |
A1C, % | 9.3 ± 1.8 | 9.1 ± 2.0 | 9.2 ± 1.9 |
(n = 72) | (n = 48) | (n = 120) | |
Living arrangements, %a | |||
Living with parents | 33 | 43 | 41 |
Not living with parents | 67b | 57 | 58 |
Vocational plans, %a | |||
Attending college | 83 | 77 | 81 |
Not attending college | 17c | 23 | 19 |
Attending college + workingd | 97 | 81 | 91 |
Data are mean ± SD unless otherwise indicated.
Questions about living arrangements and vocational plans were not asked of three HS and two HSG participants. For the HS group, living arrangements and vocational plans indicated their future plans after graduating from high school.
Indicates those planning to not to live with parents or unsure about living arrangements.
Indicates those planning to not attend college or unsure about college plans.
Indicates the percentage of participants who work (HSG) or plan to work (HS) in addition to attending college.
The PHQ-9 Modified scores for the cohort were as follows: 16% had a score of 5–9, corresponding to mild depression; 6% had a score 10–14, corresponding to moderate depression; 0.8% had a score of 15–19, corresponding to moderately severe depression, and 0.8% had a score of 20–27, corresponding to severe depression. In the overall cohort, there was no association between PHQ-9 Modified scores and A1C. However, we did find a negative association between PHQ-9 modified score and DSES and DQOL, as reported in the literature (Table 2).
. | . | DSES, Total Score . | A1C, % . | DQOL, Mean Score . | PHQ-9 Modified, Mean Score . |
---|---|---|---|---|---|
DSES, total score | Correlation coefficient | 1.00 | −0.23* | 0.51** | −0.31** |
Significance (two-tailed) | — | 0.01 | 0.000 | 0.00 | |
n | 125 | 123 | 125 | 125 | |
A1C, % | Correlation coefficient | −0.23* | 1.00 | −0.23* | 0.04 |
Significance (two-tailed) | 0.01 | — | 0.01 | 0.64 | |
n | 123 | 123 | 123 | 123 | |
DQOL, mean score | Correlation coefficient | 0.51** | −0.23* | 1.00 | −0.41** |
Significance (two-tailed) | 0.00 | 0.01 | — | 0.00 | |
n | 125 | 123 | 125 | 125 | |
PHQ-9 Modified, mean score | Correlation coefficient | −0.31** | 0.04 | −0.41** | 1.00 |
Significance (two-tailed) | 0.00 | 0.64 | 0.00 | — | |
n | 125 | 123 | 125 | 125 | |
Time since diagnosis, years | Correlation coefficient | −0.09 | 0.23* | −0.05 | −0.09 |
Significance (two-tailed) | 0.29 | 0.01 | 0.57 | 0.33 | |
n | 125 | 123 | 125 | 125 |
. | . | DSES, Total Score . | A1C, % . | DQOL, Mean Score . | PHQ-9 Modified, Mean Score . |
---|---|---|---|---|---|
DSES, total score | Correlation coefficient | 1.00 | −0.23* | 0.51** | −0.31** |
Significance (two-tailed) | — | 0.01 | 0.000 | 0.00 | |
n | 125 | 123 | 125 | 125 | |
A1C, % | Correlation coefficient | −0.23* | 1.00 | −0.23* | 0.04 |
Significance (two-tailed) | 0.01 | — | 0.01 | 0.64 | |
n | 123 | 123 | 123 | 123 | |
DQOL, mean score | Correlation coefficient | 0.51** | −0.23* | 1.00 | −0.41** |
Significance (two-tailed) | 0.00 | 0.01 | — | 0.00 | |
n | 125 | 123 | 125 | 125 | |
PHQ-9 Modified, mean score | Correlation coefficient | −0.31** | 0.04 | −0.41** | 1.00 |
Significance (two-tailed) | 0.00 | 0.64 | 0.00 | — | |
n | 125 | 123 | 125 | 125 | |
Time since diagnosis, years | Correlation coefficient | −0.09 | 0.23* | −0.05 | −0.09 |
Significance (two-tailed) | 0.29 | 0.01 | 0.57 | 0.33 | |
n | 125 | 123 | 125 | 125 |
Correlation is significant at the 0.05 level (two-tailed).
Correlation is significant at the 0.01 level (two-tailed).
Perceived DSES for the overall cohort was 8.2 ± 1.30. The average DQOL-BCI score was 74.3 ± 12.9. There were no significant differences in DSES score, DQOL-BCI score, depressive symptoms, or A1C values between the HS and HSG groups (P >0.05). The 13 participants who met ADA-recommended A1C goals had significantly higher mean DSES (9 ± 0.7, P <0.05) and DQOL-BCI (82.8 ± 9.6, P <0.05) scores and lower A1C concentrations (6.6 ± 0.5%, P <0.05) than the rest of the cohort. However, they did not differ significantly in mean depressive symptoms.
The mean DSES score for the entire cohort (8.2 ± 1.3) correlated negatively with A1C (R −0.23, P = 0.01). There was also a negative correlation between DQOL-BCI score (74.3 ± 12.9) and A1C (R −0.23, P = 0.01).
Exploratory Analyses
The exploratory analyses included two additional constructs: work/vocational plans and living situation. For the overall cohort, 81% either attended (HSG) or planned to attend (HS) college. The majority of participants (91%) who were attending or planning to attend college reported working or planning to work at the same time. Additionally, 58% of participants reported current or future living arrangements away from home. Among the HSG group, 77% were attending college, and 81% of these youth reported working either part time or full time. Fifty-seven percent of HSG participants reported that they were not living with parents. In the HS group, 83% had plans to attend college, whereas the remainder either did not plan to attend college or were not sure of their future educational plans. With regard to vocational plans for the future, 97% of the HS participants who planned to attend college also planned to work, and only 33% planned to live at home.
The association between DSES and A1C was further tested using general linear models. Living situation and work were added to the models, first as separate constructs and then as a combined moderation term accounting for participants’ living situation and work. With each of these additions, the trend between DSES and A1C was lost (Table 3). After adjustment for DSES, there was trend toward higher A1C among participants who worked and did not live with parents (Figure 1). Additionally, with the addition of living situation and work, the association between DQOL and A1C was no longer significant (data not shown). There was a significant positive correlation between DSES and DQOL (r = 0.5, P = 0.00).
Source . | Type III Sum of Squares . | Df . | Mean Square . | F . | Significance . | Partial η2 . |
---|---|---|---|---|---|---|
Corrected model | 18.05a | 4 | 4.51 | 1.24 | 0.30 | 0.05 |
Intercept | 287.40 | 1 | 287.40 | 78.69 | 0.00 | 0.46 |
DSES total | 13.35 | 1 | 13.35 | 3.66 | 0.05 | 0.04 |
Living with parents | 0.59 | 1 | 0.59 | 0.16 | 0.69 | 0.00 |
Work | 5.13 | 1 | 5.13 | 1.40 | 0.24 | 0.02 |
Living with parents * work | 0.15 | 1 | 0.15 | 0.04 | 0.84 | 0.00 |
Source . | Type III Sum of Squares . | Df . | Mean Square . | F . | Significance . | Partial η2 . |
---|---|---|---|---|---|---|
Corrected model | 18.05a | 4 | 4.51 | 1.24 | 0.30 | 0.05 |
Intercept | 287.40 | 1 | 287.40 | 78.69 | 0.00 | 0.46 |
DSES total | 13.35 | 1 | 13.35 | 3.66 | 0.05 | 0.04 |
Living with parents | 0.59 | 1 | 0.59 | 0.16 | 0.69 | 0.00 |
Work | 5.13 | 1 | 5.13 | 1.40 | 0.24 | 0.02 |
Living with parents * work | 0.15 | 1 | 0.15 | 0.04 | 0.84 | 0.00 |
Dependent variable: A1C (%).
R2 = 0.05 (adjusted R2 = 0.01).
Discussion
In this study, we evaluated patient perceptions of diabetes care self-management, QOL, and mood in a population of emerging adults with type 1 diabetes in relation to HSG. This cohort of patients with type 1 diabetes had high DSES and above average DQOL scores, and few reported depressive symptoms. However, diabetes control was poor, likely related to multiple lifestyle demands competing with type 1 diabetes self-management. The mean A1C in our cohort (9.2%) was similar to that reported by the T1D Exchange Clinic NEtwork (19), whereas the percentage of participants meeting age-appropriate ADA-recommended A1C goals (10.4%) appears to be lower (14–17%) (3,19).
Although higher scores on the DSES indicate higher perceived self-efficacy (15), these scores did not appear to translate to participants’ achievement of age-appropriate ADA-recommended A1C goals. A1C concentrations were inversely associated with DSES scores and diabetes QOL, but the associations were not strong. Nevertheless, these correlations highlight that perceived self-efficacy and above-average satisfaction with overall diabetes care per se do not always equate with optimal glycemic control. Furthermore, it is important to note that the DSES questionnaire assessed participants’ perceived self-efficacy in diabetes management, and participants were not required to demonstrate their diabetes knowledge as part of this study. We were therefore unable to evaluate any gaps in perceived versus actual knowledge.
Depression rates in our sample were similar to rates previously reported in youth with type 1 diabetes. The SEARCH for Diabetes in Youth study (20) found that 14% of youth with diabetes had mild depression, whereas 8.6% had moderate or severe depression, and Hood et al. (21) found rates of 12.8% for mild depression and 6.6% for moderate or severe depression among youth with type 1 diabetes. Although depression is associated with poor diabetes self-care, it is unlikely that depressive symptoms are the cause for poor glycemic control in our cohort because the prevalence of moderate to severe depression was very low. This finding may explain the lack of association between depression and glycemic control in our population, although we did find negative associations between DSES, DQOL, and PHQ-9 Modified scores. Other factors, including multiple lifestyle demands and/or stressors during these critical transition years, may have played a significant role in competing with diabetes self-management tasks (22,23).
In our cohort, a large percentage of participants indicated ongoing or future plans for college education along with work, while also reporting a plan to live away from home. These demographic situations could create lifestyle stressors that compete with diabetes self-care. Accordingly, we did see a trend toward higher A1C in participants who were working part time or full time and living separately from their parents. Furthermore, the inverse association between DSES and A1C could no longer be demonstrated in our participants when living arrangements and work situation were added to the equation. The current study did not evaluate diabetes distress in youth, which remains an important avenue for future research.
We recognize that our results may be limited in part by the homogeneous racial makeup of the group. At the same time, this cohort is representative of the population of patients followed at our tertiary multidisciplinary diabetes center and mirrors nationally reported data in youth with type 1 diabetes aged 18–25 years and in the SEARCH study (19,24). Of note is that the percentage of participants using an insulin pump (49.7%) for insulin delivery in our sample was slightly lower than that previously reported in this age-group (55%) (19). However, this study was not designed to examine the potential effect of modality of insulin delivery (CSII vs. MDI) on DSES, DQOL, and PHQ-9 Modified scores or A1C.
These findings underscore the ongoing need to identify the barriers between knowledge of diabetes care and implementation of such knowledge in this population. This finding is not limited to type 1 diabetes, as it has been observed in patients with obesity, prediabetes, and type 2 diabetes, for whom self-management education requires focused professional support (25). Our findings highlight an area of further study to implement transition plans by pediatric diabetes teams as they prepare older teenagers and young adults for the rigors of independence in diabetes care (26–29). Pediatric diabetes teams recognize the ADA and Endocrine Society recommendations that individuals in this age-group should receive additional diabetes knowledge and self-management education (30,31). However, despite these efforts, moving the needle toward improved glycemic control in this population remains challenging (27,32).
In conclusion, our data suggest that our cohort of young people with type 1 diabetes report good diabetes self-efficacy in comparison with mean self-efficacy scores reported by Lorig et al. (16). However, based on the lack of achievement of optimal diabetes control, high DSES scores alone may not be a positive prognostic indicator of optimal diabetes control. We recognize that a prospective design may shed additional light on the value of the parameters we measured and that the cross-sectional design of our study limits the generalization of our findings to other youth with type 1 diabetes. Future structured interventions are necessary to assist this vulnerable population as they transition into adulthood while attending college and/or entering the work force, both of which pose demands that may compete with diabetes self-management.
Article Information
Acknowledgments
The authors thank all of the participants of the study and our research study team members: H. Elsinghorst, A. House, A. Marrone, S. Michalovic, B. Shine, and E. Tabaczynski. L.D.M. receives research support from AstraZeneca, JDRF, Novo Nordisk, Sanofi Aventis. T.Q. receives support for clinical trials from Ascendis, Janssen, Opko, and Provention Bio. I.M. receives support for clinical trials from Rhythm.
Funding
Departmental funds were used to compensate study participants for their time.
Duality of Interest
No potential conflicts of interest relevant to this article were reported.
Author Contributions
All authors contributed to study design. T.S. and L.D.M. performed the literature review. T.S., L.D.M. and T.Q. planned and executed the study. T.S. collected data and performed data entry. I.M. and T.S. analyzed the data. All authors interpreted the data, participated in manuscript preparation and revisions, and approved the final manuscript: T.Q. was the principal investigator of the study and played a key role in assessing the need for the study and defining study objectives. T.Q. is the guarantor of this work and, as such, has 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.