Effects of Trauma and Anxiety on Adherence in Pediatric Type 1 Diabetes

This was a single-center, cross-sectional study. Written informed consent, or parental permission and youth assent where appropriate, was obtained before data collection. The study was approved by the University of Kentucky institutional review board, and all study activities were carried out in accordance with the Declaration of Helsinki.

Participants were 7–21 years of age and had been diagnosed with type 1 diabetes at least 6 months before enrollment, which took place during a clinic visit. Potential participants were identified and approached by their clinical care provider and then approached by the research staff to explain the study and obtain consent/assent. A total of 126 patients were approached by research staff, and, of those, 99 were recruited to participate between April 2018 and January 2020. The clinic population was used as a convenience sample; no specific populations were targeted for enrollment. Pregnancy and moderate to severe intellectual disability were exclusion criteria.

Demographic data were obtained from medical records and included participants’ age, sex, race/ethnicity, duration of diabetes, history of hospitalization for DKA in the past year, and any documented diagnosis of psychiatric or behavioral disorders (defined as any ICD-10 F-code diagnosis present in the chart).

We used point-of-care A1C measured at the time of the clinic visit as an indicator of glycemic control. Participants were considered to be at glycemic goal if their A1C was <7.0% and above goal if their A1C was ≥7.0% based on recent specialty guidelines (22,23).

As part of their clinical care, participants provided their BGM records from their glucose meter or written blood glucose records for the 2 weeks before the clinic visit. These records were scanned into the electronic health record, abstracted by one of the authors (A.L.M.) and coded based on agreement between two of the authors (R.R. and A.L.M.). In consultation with a pediatric endocrinologist (A.M.P.), behavioral adherence was coded as a dichotomous variable defined as BGM at least four times per day or, for patients using continuous glucose monitoring (CGM), calibration at least two times daily if required by the CGM system manufacturer and sensor wear at least 75% of the time averaged over 2 weeks (24). For those who did not have glucose logs or glucose meter data (n = 9), use of CGM and/or an estimate of the frequency of BGM was used based on the medical record.

We administered three standardized self-report questionnaires—the Child and Adolescent Trauma Survey (CATS), the Screen for Child Anxiety-Related Disorders (SCARED), and the Mood and Feeling Questionnaires (MFQ)—to screen for trauma exposure or post-traumatic stress symptoms, anxiety symptoms, and depression, respectively. All data were obtained via self-report.

CATS is a questionnaire to screen for traumatic exposure and PTSD symptoms based on DSM-V criteria for PTSD (25). Traumatic exposure is elicited using a 15-item structured traumatic exposure checklist. The checklist follows the definition of traumatic events in the DSM-V and includes natural disasters, accidents, experiencing or seeing violence at home or in the community, sexual abuse, traumatic loss, medical procedures, and war. If at least one traumatic exposure was present, post-traumatic stress symptoms were measured by 25 items rated on a four-point Likert scale (0 = never, 1 = once in a while, 2 = half the time, and 3 = almost always). Traumatic exposure, an aggregate post-traumatic stress symptom score >16 on the CATS screen, and at least one area of functional impairment were considered to be a positive indicator of PTSD (26).

SCARED is a screening tool for anxiety disorders in children and young adults. It is a 41-item structured questionnaire rated on a three-point Likert scale (0 = not true or hardly ever true, 1 = somewhat true or sometimes true, and 2 = very true or often true). Responses are scored by adding the point values of responses to each item. A total score ≥25 was considered positive for anxiety disorder. Scores of ≥7 on panic symptoms, ≥9 on generalized anxiety symptoms, ≥5 on separation anxiety symptoms, ≥8 on social anxiety disorder symptoms, and ≥3 on significant school avoidance symptoms were considered positive for panic disorder, generalized anxiety disorder, separation anxiety, social anxiety disorder, and significant school avoidance, respectively (27).

The MFQ is a screening tool for depression in children and young adults that is validated for use with patients up to 7 years of age. It consists of a series of 33 descriptive phrases and is rated on a three-point Likert scale (0 = not true, 1 = somewhat true, and 2 = true) and scored by adding the point values of responses for each item. A total score of ≥29 was considered positive for depression, and a score ≥1 on any of the four questions related to suicide was considered positive (28,29). Although there are no established score cutoffs for the MFQ, previously published research has determined that a score of 29 optimally discriminates youth with major depressive episodes from others (30). The four items about suicidal ideation in the previous 2 weeks are: “I thought that life wasn’t worth living,” “I thought about death or dying,” “I thought my family would be better off without me,” and “I thought about killing myself.” We considered an answer of “somewhat true” or “true” on any of the four questions as a marker of suicidal ideation. Answers of “somewhat true” or “true” on the fourth question (“I thought about killing myself”) in those who had not already had a safety evaluation during the clinic visit were referred for additional safety assessment in the clinic.

After obtaining consent/assent, study personnel approached participants and their guardians to complete questionnaires. Responses to the questionnaires were collected on iPads and managed via a secure web-based app called REDCap (Research Electronic Data Capture) (31) hosted at the University of Kentucky. The total time to complete questionnaires was 15–20 minutes. Participants completed the questionnaires in the same clinic room with their parent or guardian unless there was a clinical indication to interview participants alone. Children who disclosed ongoing or unreported abuse or suicidal ideation were interviewed by the clinic social worker or a child and adolescent psychiatrist associated with the study, as described in results.

We analyzed data using SPSS v.25 software (SPSS Inc., Chicago, IL). Descriptive analyses were conducted. Anxiety, suicidal ideation, trauma exposure, and post-traumatic stress symptoms were coded as dichotomous variables according to their clinical cutoffs as described above. A1C was a continuous variable. χ² and Fisher exact tests were used to assess the relationships among post-traumatic stress symptoms, anxiety, suicidal ideation, and behavioral adherence. φ Coefficients were used to assess the strength of statistically significant relationships between the paired categorical variables. One-way ANOVA was conducted to examine group differences between A1C and presence of suicidal ideation. Levene’s test was used to assess homogeneity of variances.

The datasets generated and/or analyzed during the study are not publicly available to ensure patient privacy but may be available from the corresponding author upon reasonable request.

Demographic and clinical characteristics of the participants are summarized in Table 1. The study included 99 participants, and 76.1% (99/126) of patients who were approached by research staff agreed to participate. No data were collected on those who were approached but declined to participate. The average age of the study population was 13.8 ± 3.5 years, and 51.5% of participants were >13 years of age. The majority of study participants were White (96%), with 3% African American; 3% reported that they were Hispanic or Latino, and 93% were non-Hispanic/Latino. The average A1C was 8.4 ± 1.4%, the average duration of type 1 diabetes was 4.6 ± 3.2 years, 6.1% of participants had been hospitalized for DKA in the past year, and 18.2% had at least one previously diagnosed psychiatric disorder documented in the medical chart.

Among the 65.7% of study participants who were exposed to traumatic events, 38.4% also had post-traumatic stress symptoms and functional impairment concerning for PTSD, whereas 30.6% screened positive for symptoms of anxiety. We also examined rates of depression and suicidal ideation, which were 7.1 and 25.5%, respectively. These findings are summarized in Table 2. Thirteen percent of participants required additional safety evaluation by the clinic social worker based on their answers to the questionnaires, which occurred in the clinic, except for one patient who had left the clinic before the questionnaire answers were reviewed and had a safety assessment conducted over the telephone on the same day as the appointment. If the social worker was unavailable, one of the authors (A.L.M.), a child and adolescent psychiatrist, conducted the safety assessment. No participants reported active suicidal intent or plan.

Forty percent of youth met the study’s definition of behavioral adherence (encompassing a specific set of self-management behaviors focusing on frequency of BGM and calibration of CGM system and/or wearing of a CGM sensor (see research design and methods). Post-traumatic stress symptoms with functional impairment were not associated with behavioral nonadherence (P = 0.546). Although overall anxiety disorders and depression were not associated with behavioral nonadherence (P = 0.337 and P = 0.697, respectively), the subscales for significant school avoidance and generalized anxiety disorders were associated with behavioral nonadherence (P = 0.023 and P = 0.032, respectively). These findings are summarized in Table 3. There was no association between trauma and hospitalization for DKA in the past year (P = 0.177), between reaching A1C goal and chart diagnosis of a psychiatric or behavioral concern (P = 0.683), or between chart diagnosis of a psychiatric or behavioral concern and nonadherence (P = 0.741). Furthermore, participants with the presence of suicidal ideation had a higher mean A1C than without (8.9 vs. 8.3%, P = 0.047). A Cook’s distance plot and Q-Q plot both indicated there were no outliers that affected the normality assumption. Additionally, there was no association between PTSD and having A1C at goal (P = 0.745).

Our study found a high degree of exposure to potentially traumatic events for our pediatric diabetes population, with nearly two-thirds of participants reporting at least one traumatic event, similar to but slightly higher than rates reported from community samples. The most common trauma exposure was accidental injury, with medical traumatic stress as the second most common. Although we did not require participants to describe their traumatic experiences, prior work has established that many children with diabetes find the experiences related to their diagnosis traumatic (21). The screening questionnaires we used did ask questions about some experiences that could be considered adverse childhood experiences, including physical and sexual abuse, as well as witnessing family violence; however, the CATS trauma screening tool we used does not specifically ask about emotional abuse or neglect or experiences related to household dysfunction, nor does it calculate a score related to traumatic exposure, which may account for differences between our study and that of Iqbal et al. (14). Additionally, we did not examine the timing of traumatic events, which may account for differences from other studies that have examined past-year stressful life events (13,32).

Despite recognition that childhood traumatic experiences are important factors in later health outcomes, the behavioral and metabolic changes that occur as a result of those experiences are poorly understood (33,34). Our study examined potential relationships among exposures to trauma, post-traumatic stress symptoms, and diabetes self-management. We found no relationship between trauma exposure or symptoms of PTSD and either diabetes self-management or glycemic control.

Our study found that nearly one in three participants reported some degree of suicidal ideation in the past 2 weeks. Of note, the rate of clinically significant depression symptoms was lower than the rate of suicidal ideation. Given that the presence of suicidal ideation was associated with worse glycemic control, it may be important for diabetes care providers to specifically screen for suicidality in addition to screening for depression symptoms. Additionally, we did replicate previous work that established a relationship between anxiety and worse behavioral adherence, highlighting that both generalized anxiety and school avoidance may be important contributors (35).

Psychiatric and psychological factors remain an important contributor to BGM adherence for children, adolescents, and young adults. Our study reaffirmed that anxiety and suicidal ideation may be particularly important factors in diabetes self-management. Many diabetes centers have access to multidisciplinary teams, which can include social work, psychology, or psychiatry experts, to implement routine screening, refer patients for treatment, and coordinate care with endocrinologists and other diabetes specialists. Understanding the psychiatric factors that may contribute to glycemic control and diabetes self-management will allow more effective education, screening, and interventions, especially early in the course of pediatric diabetes. Education for patients and families should include information about the psychological factors that may play a role in diabetes control, including anxiety and suicidal ideation. However, there can also be reassurance that exposure to stressful events may not always lead to difficulties with adherence or glycemic control and that recent stressful events and more severe traumatic exposures such as adverse childhood experiences may play a more important role.

As with all studies, there are some limitations to be noted. This was a retrospective, cross-sectional analysis that could not determine directionality of associations. The sample was primarily non-Hispanic White, which, although reflective of the local population and catchment area of the clinic, may affect generalizability to other populations. We included a wide range of ages, which may have resulted in variable diabetes self-management as well as differential traumatic exposure and experience. Our sample was a clinic sample, with some candidates refusing to participate in the study, although we did not characterize those who declined to participate or those who were not approached for the research. We also relied on children’s reports of traumatic events and their perception of the impact on their functioning, which may have been limited by recall and subjective interpretation of impairment. Finally, there are some methodological differences in how trauma exposure was defined in our study compared with prior work (14), which may have influenced the reporting and outcomes.

Our study did not find an association between potentially traumatic events or post-traumatic symptoms and behavioral adherence to BGM or glycemic control as measured by A1C. However, we did reaffirm the importance of anxiety and depression, and especially suicidal ideation, as potential factors in self-management and glycemic control. Medical providers who care for children, adolescents, and young adults with type 1 diabetes should ask about stressful life experiences and psychiatric symptoms and refer youth for appropriate interventions to optimize diabetes care. Future studies may investigate the timing of traumatic experiences and the degree of symptom burden to clarify the relationship between trauma and glycemic control.

This project was supported by a Research Starter Grant from the Kentucky Children’s Hospital–Children’s Miracle Network Research Fund and the National Institutes of Health (NIH) National Center for Advancing Translational Sciences, through grant number UL1TR001998. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

No potential conflicts of interest relevant to this article were reported.

A.M.P., M.L.M., and A.L.M. conceptualized and designed the study. R.R., M.N., A.M.P., M.L.M., and A.L.M. collected the data. M.N., O.V., and A.L.M. analyzed the data. R.R., M.N., and A.L.M. created the first draft of the manuscript. All authors contributed to interpretation of the data, critical revision of the manuscript, and final approval of the version to be published. A.L.M. 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.

An earlier version of this work was presented as a poster at the American Diabetes Association’s virtual 80th Scientific Session in June 2020.