Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jul 1.
Published in final edited form as: Can J Diabetes. 2021 May 11;45(5):473–480. doi: 10.1016/j.jcjd.2021.05.002

Examining the Indirect Effects of Anxiety on Glycated Hemoglobin via Automatic Negative Thinking and Diabetes-specific Distress in Adolescents with Type 1 Diabetes

Anthony T Vesco a,b, Kelsey R Howard a, Lindsay M Anderson c, Jaclyn L Papadakis a,b, Korey K Hood d, Jill Weissberg-Benchell a,b
PMCID: PMC8239251  NIHMSID: NIHMS1702991  PMID: 34176611

Abstract

Objectives:

This study examined the indirect effects of anxiety on A1c via automatic negative thinking and diabetes distress among adolescents with type 1 diabetes (T1D) during the follow-up interval of a randomized controlled trial of an intervention targeting resilience promotion/depression prevention.

Methods:

Adolescents (N=264) participating in the Supporting Teen Problem Solving (STePS) clinical trial were included and assessed at 8-, 12-, 16-, and 28-months post-baseline. A serial, double mediation model was used to examine path effects from anxiety to A1c through automatic negative thinking, diabetes distress, and through both automatic negative thinking and diabetes distress. Relevant demographic and clinical covariates were included.

Results:

Anxiety significantly predicted increases in both automatic negative thinking and diabetes distress. Automatic negative thinking was not found to mediate the association between anxiety and A1c but diabetes distress did mediate the association. The double mediation path through automatic negative thinking and diabetes distress together was significant. The indirect effect of anxiety on A1c through diabetes distress was significant and greater than the indirect effect of the double mediator path. Anxiety did not predict A1c independent of its effects on automatic negative thinking and diabetes distress. Inclusion of demographic covariates did not substantively change results.

Conclusions:

Analyses suggest that automatic negative thinking and diabetes distress mediate the relationship between anxiety and A1c among adolescents with T1D. Diabetes distress appears to be a robust factor linking anxiety to A1c. Diabetes distress should be further examined as a mediator of glycemic variability in anxious youth with T1D.

Keywords: type 1 diabetes, diabetes distress, A1c, automatic negative thinking, adolescents, mediator

Introduction

Anxiety is defined as being emotionally hypersensitive to one’s environment and demonstrating greater worry, physiological arousal, and concentration difficulties [1]. Anxiety can predispose youth with type 1 diabetes (T1D) to greater emotional distress than typically expected from daily T1D burdens and may impact diabetes care. Prevalence of elevated anxiety symptoms in youth with T1D ranges from 13 – 21% [2,3], and about 18% of youth with T1D will be diagnosed with an anxiety disorder before adulthood [4]. Anxiety is associated with higher A1c for youth with T1D and has demonstrated longitudinal associations with A1c up to one year later [5,6]. Although this association is known, it is unclear what is driving, or mediating, this link. Fisher and colleagues described a pathway, in adults with T1D, linking poor emotion regulation (which may include anxiety) to increased diabetes distress to worse glycemic outcomes [7]. This model demonstrates the broad influence of poor emotion regulation on diabetes distress and glycemic variability [8]. However, the pathway linking poor emotion regulation to diabetes distress to glycemic variability has not been investigated in youth with T1D.

Poor emotion regulation is associated with negative self-evaluation, worry, and automatic negative thinking. Automatic negative thinking refers to the pessimistic attitudes held about self, others, and the world, and associated perceived meaning and implications, that is associated with negative mood states [9,10]. Automatic negative thinking is a transdiagnostic process associated with both anxiety and depression. It differentiates people with clinical elevations in either depression or anxiety from people with neither, although certain types of automatic thoughts are more associated with depression than anxiety [10]. Increased anxiety can cause youth to assume neutral or ambiguous stimuli are dangerous and avoid such stimuli [11]. Increased anxiety is often associated with increased negative-self talk [12], likely stemming from automatic negative thinking.

For adolescents with T1D, and in line with Fisher and colleagues’ model, elevated anxiety and associated automatic negative thinking may contribute to increased diabetes distress. Diabetes distress refers to the negative emotions associated with the daily frustrations of diabetes management and perceptions of not being fully supported by family or friends with diabetes [13–15]. While automatic negative thinking is a construct that taps generalized thought perseveration and content related to negative mood states, diabetes distress taps negative feelings and thoughts specific to constant T1D demands. In adolescents with T1D, diabetes distress is associated with suboptimal diabetes management and increased glycemic variability [16]. Using Fisher and colleagues’ work as a basis, it is plausible that anxiety may contribute to increased automatic negative thinking and diabetes distress in adolescents with T1D which, in turn, may be associated with increased glycemic variability. Critical judgments about emotions, emotional reactivity, and worrying are associated with higher diabetes distress in adults with T1D [7]. Such processes may also occur in adolescents with T1D. Further investigation of diabetes distress as a mediator of the anxiety-A1c link in adolescents with T1D is warranted.

Both automatic negative thinking and diabetes distress could be treatment targets for psychological and behavioral interventions for adolescents with T1D. Understanding the roles of these mediators can inform clinical approaches to reducing the effect of anxiety on glycemic variability in youth with T1D. Psychologically-based interventions show promise in ameliorating anxiety, diabetes distress, and lowering A1c in adults, especially when general interventions are tailored for diabetes [17–20]. Some behavioral interventions conducted in youth demonstrate promise in reducing anxiety [21,22], diabetes distress [23,24], and glycemic variability [25]. Clarifying potential mechanisms of change for anxiety, diabetes distress, and glycemic variability in youth with diabetes would inform the developing treatment literature. Further, studying the role of automatic negative thinking offers an opportunity to explore additional treatment targets.

Several demographic and diabetes-related factors may drive the association between anxiety and A1c in adolescents with T1D. Among youth with T1D, girls tend to report higher anxiety and worry than boys [26–28]. Association between use of diabetes technology (i.e., using continuous glucose monitoring [CGM] and/or insulin pumps) and anxiety is mixed, with some studies reporting decreased anxiety with technology use and others with unintended anxiety related to raising awareness to glycemic trends otherwise unknown [29]. No other clear demographic associations for anxiety in youth with T1D are documented. For A1c, older adolescent age and being a member of a non-White racial group predict higher A1c and a stable high trajectory [30–32]. Two-parent households, higher family income, and use of insulin pumps and CGM are associated with lower A1c [30,33–35]. Inclusion of these factors in models examining adjusted effects of automatic negative thinking and diabetes distress on the anxiety A1c link assists in determining robustness of mediation effects.

The present study examined how anxiety indirectly influences A1c through automatic negative thinking and diabetes distress in adolescents with T1D using longitudinal, follow-up data of a randomized controlled trial (RCT). Anxiety at an initial timepoint was hypothesized to influence automatic negative thinking and diabetes distress which sequentially influence A1c. Longitudinal mediation analysis with each construct measured at consecutive timepoints demonstrated causal order of the mediation model. Further, based on the extant work of Fisher and colleagues [7], it was surmised that anxiety first influences automatic negative thinking and then diabetes distress (i.e., serial, double mediation depicted in Figure 1). The authors hypothesized that indirect effects of diabetes distress and automatic negative thinking would be statistically significant in addition to the indirect effect of the double mediation. Examination of these mediators, while accounting for demographic and diabetes-related covariates, allows for a fuller understanding of how anxiety negatively influences A1c which can inform treatment design and implementation. RCT intervention was not hypothesized to influence the mediation analyses, but randomization group differences were examined.

Figure 1:

Figure 1:

Open in a new tab

Schematic diagram of proposed mediation model. Indirect effects of anxiety on A1c are being modeled through a double, serial mediation path (path effect abc) in addition to single mediation paths through automatic negative thinking (path effect ae) and diabetes distress (path effect dc). Follow-up assessment refers to the number of months following baseline assessment in the larger study from which data originate. Measures associated with constructs appear in parentheses. STAI=State-Trait Anxiety Inventory; ATQ=Automatic Thoughts Questionnaire; PAID-T=Problem Areas in Diabetes, Teen Self-report.

Methods

Participants

This study included adolescents with T1D who participated in the Supporting Teen Problem Solving (STePS) clinical trial. All participants regardless of randomization condition were included. Study recruitment methods, inclusion and exclusion criteria, and enrollment procedures are reported elsewhere [36]. In total, 264 adolescents participated in the STePS clinical trial. Retention for this study was 96%.

Procedure

Study design and method for STePS are reported elsewhere [36,37] and were approved by institutional human subjects review boards at participating sites. In brief, STePS was an RCT evaluating the efficacy of a depression-prevention/resilience-promoting group therapy program for youth with T1D. Participants were consented then randomized into either therapy or an educational program control. Data were collected at baseline, post-intervention, and at 8-, 12-, 16-, and 28-months post-baseline. For the current mediation analyses, anxiety at 8-month follow-up, automatic negative thinking at 12 months, diabetes distress at 16 months, and A1c at 28 months were used. This manipulated the causal order of the variables to be in line with the proposed longitudinal mediation model (see Figure 1).

Measures

Anxiety.

The 20-item trait subscale of the State-Trait Anxiety Inventory (STAI) [38] was used to measure anxiety. The STAI is a reliable measure of both state anxiety (i.e., how worried, tense, or nervous participants feel in the moment) and trait anxiety (i.e., how worried, tense, or nervous participants are in general). Adolescents rate each item on a 4-point scale, with 1=“Almost Never” and 4=“ Almost Always.” Higher scores on the trait subscale correspond with higher anxiety. Reliability for the STAI is excellent, with internal consistency ranging from ɑ=0.86-0.95 [38].

Automatic negative thinking.

The Automatic Thoughts Questionnaire (ATQ) [39] was used to measure severity of automatic negative thinking. The ATQ is a 30-item questionnaire that measures the frequency of negative self-statements over the past week. Sample items include “My life’s not going the way I want it to” and “I’m a failure.” Items are rated on a 5-point scale from 1=“Not at All” to 5=“ All the Time.” A sum score is generated with higher scores indicating more frequent automatic negative thinking. The reliability for the ATQ is excellent [39].

Diabetes distress.

Diabetes distress was measured using the Problem Areas in Diabetes, Teen Self-Report (PAID-T) [13]. The PAID-T is a 14-item questionnaire assessing diabetes distress in adolescents over the past month. Sample items include “Feeling sad when I think about having and living with diabetes;” “Feeling like my parents don’t trust me to care for my diabetes;” and “Feeling that I am not checking my blood sugars often enough.” Items are rated on a 6-point scale from 1=“Not a Problem” to 6=“Serious Problem.” Higher total scores indicate greater distress. Reliability for the PAID-T is strong, with internal consistency ranging from α=0.85-0.93 for individual subscales [13].

A1c.

Hemoglobin A1c is a measure of the percent of glycated hemoglobin in the blood which serves as a measure of glycemic variability. A1c is a proxy for the average estimate of blood glucose values over the previous 3 months. Adolescents provided a small sample of capillary blood obtained during each assessment with a kit provided by a central laboratory. The laboratory then tested the sample to obtain the A1c value.

Statistical Analyses

Analysis key outcomes were anxiety, automatic negative thinking, diabetes distress, and A1c. Descriptive statistics and correlations of these variables and participant demographic and diabetes-related characteristics were tabulated. Correlations and t-tests were used to determine the extent of statistical association between these characteristics and each of the key outcomes. Cohen’s d effect sizes and associated confidence intervals were calculated for any binary characteristics and presented along with t-test significance to show strength of association.

A serial, double mediation model was used to investigate the indirect effects of anxiety on A1c through automatic negative thinking and diabetes distress (see Figure 1). Serial, double mediation allows for the path effects from anxiety to A1c through automatic negative thinking, through diabetes distress, and through both automatic negative thinking and diabetes distress to be estimated simultaneously in a single model. For these analyses, maximum likelihood estimation was used, and bias-corrected, bootstrap confidence intervals for all effects were estimated based on 10,000 bootstrap samples. Indirect effects for the proposed mediation were calculated, and statistical significance was determined if zero fell outside the estimated confidence interval [40]. A separate mediation model, of the same form as Figure 1 but including statistically relevant covariates determined via preliminary analyses of participant characteristics with key outcomes, was also estimated to determine robustness of indirect effects of proposed mediation paths. Categorical variables were dichotomized and dummy-coded prior to analyses to allow for appropriate effect estimation. Pairwise differences in indirect effects among the double mediation path and anxiety and diabetes distress each alone as mediators were completed by examining the difference in indirect effect product coefficients and constructing a bias-corrected, bootstrapped confidence interval around this difference. Like the other significance tests, if zero fell outside of this confidence interval, then the difference was deemed to be statistically significant.

Given that these analyses used the follow-up data of an RCT, the association of the randomization group with each of the key outcomes was also assessed using t-tests. Further, examination of the effect of randomization group on indirect effects in each mediation analysis was completed by estimating indirect effects for each group using a multigroup path analysis. Calculated indirect effects were compared between groups using the difference estimate and calculated confidence interval around the difference in the same manner as noted above. All mediation analyses were conducted using the lavaan package in R version 3.6.2 [41,42].

Results

Participant Characteristics

Table 1 presents a summary of participant demographic and diabetes-related information. The sample was composed of older adolescents who were aged 14–19 years at the 8-month follow-up visit. Adolescents were living with diabetes for seven years on average (range: 1.6–16.5 years). The majority of the sample was composed of White girls, although nearly one-third of the sample was non-White (34.5%). Most of the sample was using an insulin pump (68.2%) and a small portion was using CGM (28.5%). Many adolescents were living with both of their biological parents (70.5%). The median annual family income was between $100-125,000 USD. For all further analyses, the categorical family income variable was dichotomized to be either below or equal to the median or above the median (i.e., ≤$100,000 and >$100,000 USD). Also, given the low sample sizes for non-White racial groups, this variable was also dichotomized into two groups: non-White and White.

Table 1:

Participant characteristics (N=264)

Characteristic Mean SD
Age (years) 16.5 1.1
Duration of T1D (years) 7.7 4.0
N %
Gender Male 106 40.2
Female 158 59.9
Race/Ethnicity Black or African American 38 14.4
Hispanic/Latinx 29 11.0
Native American or Alaska Native 3 1.1
Asian or Pacific Islander 6 2.3
White/European American 173 65.5
Other Identified Race 15 5.7
Mode of Insulin Delivery Pump 161 68.2
Injections 75 31.8
Mode of Blood Glucose Checking CGM 67 28.5
SBGM 168 71.5
Family Composition Living with two biological parents 167 70.5
Single or blended family 70 29.5
Annual Family Income* ≤ $25,000 11 5.1
$26 - 50,000 16 7.4
$51 - 75,000 34 15.8
$76 - 100,000 30 14.0
$101 - 125,000 27 12.6
$126 - 150,000 21 9.8
$151 - 175,000 15 7.0
> $175,000 38 17.7
Open in a new tab

Note:

*

All monetary amounts are presented in USD.

SD=Standard Deviation; T1D=Type 1 diabetes; CGM=Continuous Glucose Monitoring; SBGM=Standard Blood Glucose Monitoring

Associations of Outcomes with Participant Characteristics

Table 2 provides descriptive statistics and correlations of the primary outcomes (STAI, ATQ, PAID-T, and A1C). Examination of differences between participant characteristics and outcomes of interest yielded several statistically significant associations. For the STAI, girls had higher scores than boys (p=0.003, d=0.39, 95% CI: 0.13-0.66). For the ATQ, standard blood glucose monitoring (SBGM) use was associated with higher scores than CGM use (p=0.025, d=0.33, 95% CI: 0.04-0.62). For the PAID-T, girls had higher scores than boys (p=0.001, d=0.45, 95% CI: 0.18-0.73), and non-White adolescents had higher scores than White adolescents (p=0.046, d=0.29, 95% CI: 0.01-0.57). For A1c, non-White racial group status was associated with higher values compared to White youth (p<0.001. d=0.74, 95% CI: 0.44-1.03); injection use was associated with higher values compared to pump use (p=0.002, d=0.47, 95% CI: 0.18-0.77); being in a single or blended family was associated with higher values compared to being in a two biological-parent home (p=0.033, d=0.33, 95% CI: 0.03-0.63); and lower family income was associated with higher values compared to higher family income (p=0.006. d=0.42, 95% CI: 0.12-0.72). Adolescent age and duration of living with T1D were not significantly associated with outcomes of interest (p’s>0.05).

Table 2:

Descriptive statistics and correlations of primary outcomes

Outcome Mean SD Range Correlations
STAI-Trait ATQ PAID-T A1c Age T1D Duration
STAI-Trait 37.4 11.9 (20 - 79) --- 0.66 0.58 0.26 −0.08 −0.05
ATQ 47.7 24.5 (30-148) --- --- 0.48 0.12 −0.02 −0.09
PAID-T 35.9 15.9 (14 - 79) --- --- --- 0.33 <0.01 <0.01
A1c 9.3 2.1 (5.2 - 17.2) --- --- --- --- 0.02 <0.01
Open in a new tab

Note: Correlations in bold typeface are statistically significant (p<0.05).

SD=Standard Deviation; STAI=State-Trait Anxiety Inventory; ATQ=Automatic Thoughts Questionnaire; PAID-T=Problem Areas in Diabetes, Teen Self-report.

Automatic Negative Thinking and Diabetes Distress as Mediators of the Anxiety-A1c Association

A serial, double mediation model was fit to the longitudinal data to investigate automatic negative thinking and diabetes distress as mediators of the association between anxiety and A1c. Figure 1 depicts the model, and the top half of Table 3 presents the results of this mediation analysis. Anxiety was demonstrated to significantly predict increased automatic negative thinking and diabetes distress. A1c was predicted by diabetes distress but not by automatic negative thinking. The bias-corrected bootstrap confidence interval for the indirect effect of automatic negative thinking as a single mediator contained zero, signifying that anxiety does not influence A1c via automatic negative thinking alone. Conversely, the confidence interval for the indirect effect of diabetes distress as a single mediator was entirely above zero indicating anxiety influences A1c indirectly via diabetes distress. The indirect effect of the serial, double mediation pathway through automatic negative thinking and diabetes distress had a confidence interval entirely above zero indicating that double mediation was founded. Further, the indirect effect of anxiety on A1c through diabetes distress was significantly greater than the indirect effect through both automatic negative thinking and diabetes distress (95% CI of the difference in effects was 0.003 – 0.031). No other differences between indirect effects were remarkable. There was no evidence that anxiety influenced A1c independent of its effects on automatic negative thinking and on diabetes distress (i.e., c’ was non-significant).

Table 3:

Results of mediation analyses

Mediation Model Outcome Predictor (Parameter) Estimate SE 95% CI
No covariates ATQ STAI Trait (a) 1.401 0.133 1.138 - 1.662
PAID-T STAI Trait (d) 0.598 0.107 0.377 - 0.799
ATQ (b) 0.141 0.058 0.032 - 0.257
A1c STAI Trait (c’) 0.010 0.017 −0.021 - 0.045
ATQ (e) −0.005 0.008 −0.021 - 0.012
PAID-T (c) 0.038 0.009 0.021 - 0.055
Indirect Effects of STAI on A1c Double Mediation Path (abc) 0.007 0.004 0.002 - 0.016
ATQ Mediation Path (ae) −0.006 0.012 −0.029 - 0.017
PAID-T Mediation Path (dc) 0.023 0.006 0.012 - 0.037
With covariates ATQ STAI Trait (a) 1.442 0.152 1.124 - 1.723
CGM Use −4.043 3.120 −9.843 - 2.294
PAID-T STAI Trait (d) 0.453 0.118 0.211 - 0.680
ATQ (b) 0.199 0.059 0.085 - 0.319
Gender: Female 3.647 2.189 −0.612 - 7.898
Racial Status: Non-White 4.771 2.101 0.792 - 9.062
A1c STAI Trait (c’) −0.008 0.017 −0.040 - 0.027
ATQ (e) 0.008 0.009 −0.011 - 0.026
PAID-T (c) 0.032 0.011 0.010 - 0.052
Racial Status: Non-White 1.064 0.335 0.430 - 1.741
Insulin Delivery: Pump −0.460 0.343 −1.142 - 0.212
Family Composition: 2 Biological Parents 0.161 0.351 −0.538 - 0.846
Annual Family Income: > $100,000 USD −0.732 0.291 −1.331 - −0.183
Indirect Effects of STAI on A1c Double Mediation Path (abc) 0.009 0.004 0.003 - 0.019
ATQ Mediation Path (ae) 0.011 0.014 −0.016 - 0.037
PAID-T Mediation Path (dc) 0.014 0.006 0.004 - 0.029
Open in a new tab

Note: Parameter estimates are all in unstandardized form. Coefficient estimates in bold typeface are statistically significant (p<0.05). Parameter labels in parentheses refer to the path coefficients depicted in Figure 1. Confidence intervals were calculated via bias-corrected bootstrapping of 10,000 samples.

SE=Standard Error; 95% CI=95% Confidence Interval; STAI=State-Trait Anxiety Inventory; ATQ=Automatic Thoughts Questionnaire; PAID-T=Problem Areas in Diabetes, Teen Self-report.

The same model was re-run, and any participant characteristics found to be significantly associated with a key outcome were included as covariates. In this model, CGM use (compared to SBGM) was included as a predictor of ATQ; being female (versus male) and racial group status as non-White (versus being White) were included as predictors of PAID-T; and racial group status as non-White (versus being White), family being composed of two biological parents (versus single/blended home), and family income > $100,000 USD (versus ≤ $100,000 USD) were included as predictors of A1c. Results are presented in the bottom half of Table 3. Including covariates in the model did not substantively change the magnitude of any of the estimated effects. Anxiety still demonstrated an indirect effect on A1c through automatic negative thinking and diabetes distress together and through diabetes distress alone. Notably, none of the pairwise comparisons of the indirect effects were statistically different from one another in this model. Like the original model, there was no evidence that anxiety influenced A1c independent of its effects on automatic negative thinking and diabetes distress.

Given that these analyses were conducted in the follow-up interval of an RCT, the associations of randomization assignment on key outcomes were examined. There were no randomization group differences on STAI, ATQ, PAID-T, or A1c measured at each of the respective time points. Examination of randomization group differences on indirect effects of the two mediation models (general and with covariates) yielded no statistically significant group differences. Together, there was no evidence that the intervention influenced mediation effects.

Discussion

The present study examined the indirect influence of anxiety on A1c through automatic negative thinking and diabetes distress in adolescents with T1D. The authors hypothesized that automatic negative thinking and diabetes distress would mediate the association of anxiety with A1c, both as independent mediators and through double, serial mediation. Results indicated that while anxiety alone was associated with subsequent A1c, automatic negative thinking and diabetes distress mediated this relationship. Specifically, the double mediation of anxiety on A1c via automatic negative thinking and diabetes distress was statistically meaningful. Automatic negative thinking as a single mediator was not statistically significant while diabetes distress was and had a greater indirect effect than automatic negative thinking and diabetes distress in double mediation. These relationships were robust even in the presence of significant demographic and diabetes-related covariates.

Study results are consistent with findings in the adult diabetes literature citing associations between both general psychological distress [43] and diabetes distress with A1c [44,45]. Moreover, the current study’s findings are in line with previous work conducted by Fisher and colleagues [7]. As previously noted, this group developed and validated a model of diabetes distress as a mediating influence on the link between poor emotion regulation/thought rumination and glycemic outcomes in adults with T1D. The current work extends and validates this model in adolescents with T1D. This study highlights the influence of psychosocial factors on glycemic outcomes and the explanatory role of diabetes distress in linking anxiety with A1c. While automatic negative thinking does play a role in transmitting the effect of anxiety on diabetes distress, our findings suggest that automatic negative thinking alone does not explain the effect of anxiety on A1c. This effect may be more pronounced in a sample with clinically elevated anxiety where automatic negative thinking is pronounced.

These results have important clinical implications. For adolescents with T1D, making diabetes distress reduction a primary target will likely be associated with less glycemic variability and may break the link between anxiety and A1c in this population. Similar findings have been shown in adults [17–20]. While there may be utility in targeting automatic negative thinking in anxious youth with type 1 diabetes, it should be targeted secondarily to diabetes distress. For youth with anxiety and without diabetes distress, addressing automatic negative thinking along with other evidence-based strategies (e.g., exposures) may be beneficial.

Limitations of the current study should be noted. First, the larger study from which these data originate excluded adolescents with a diagnosis of depression or elevated depressive symptoms. Results may differ in a clinically depressed adolescent sample. Second, the larger study also excluded participants who were not fluent in English which may limit the generalizability of findings to non-English speaking groups. Third, due to low subsample sizes of racially diverse groups, non-White racial groups were collapsed into one group for analyses. This fails to represent the varied experiences within each of these groups. Studies of the mediation models of the present analyses in each group are warranted to better understand how associations may differ for different racial groups. Fourth, the data stem from the follow-up interval of an RCT where half of participants received a psychological intervention. While results of the present analyses show that randomization assignment was not statistically associated with the outcomes evaluated in this study (including the indirect effects of the mediation model), there could be variation from a purely naturalistic study in the associations of anxiety, automatic negative thinking, diabetes distress, and A1c. In previous findings [36], the study intervention was found to positively impact diabetes distress but did not demonstrate significant effects on automatic negative thinking or A1c over the course of one-year post-intervention. Further, all study outcomes demonstrated a broad distribution of values and no apparent dampening due to intervention effects. For these reasons, treatment of the data as pseudo-naturalistic is statistically plausible. Lastly, this study used A1c to measure glycemic variability. Recent clinical and research work now touts other measures of glycemic variability (e.g., time-in-range) as being a more accurate measure of this construct. Unfortunately, not many participants in this sample were using CGM which did not allow for broad calculation of another measure of glycemic variability.

Several psychosocial factors were not explored in the current study that may be related to anxiety, diabetes distress, and glycemic outcomes among youth with T1D. These include family conflict, diabetes-related strengths, and peer support and communication around diabetes. Future studies should focus on continuing to develop models of the psychosocial factors that impact glycemic outcomes, along with interactions between these variables and key outcomes of the present study. Further, validating the mediating effect of anxiety on A1c via automatic negative thinking and diabetes distress among more diverse samples of youth with respect to socioeconomic status, family structure, and culture is warranted to understand the lived experiences of all youth managing T1D.

Despite study limitations, there are many strengths of this study. First, the study had a large initial sample size (N=264) and high retention across the longitudinal follow-up interval (96%). The available data provided enough power to examine indirect effects of anxiety on A1c through two serial mediators. Second, due to the large sample, analyses accounted for key demographic and diabetes-related characteristics associated with anxiety, A1c, diabetes distress, and automatic negative thinking in adolescents with T1D, which demonstrated the robustness of the indirect effects. For both reasons, these models are advantageous as they allow for consideration of multiple factors influencing glycemic variability which is more generalizable to individuals with T1D. Moreover, these analyses enabled the investigators to demonstrate the nuances of relationships between psychological variables and glycemic variability. Lastly, this study allows for further understanding of the links between psychological wellbeing and glycemic outcomes. With this information, interventions for managing anxiety and glycemic variability in adolescents with T1D can be designed to target diabetes distress more so than automatic negative thinking. This focus when designing psychosocial treatment plans will likely lead to greater intervention effectiveness and efficiency.

Conclusion

Diabetes distress appears to be a robust and significant factor contributing to glycemic outcomes for adolescents with T1D and indirectly explains the link between anxiety and A1c. Automatic negative thinking has a small additive contribution, although not statistically significant. Future studies should continue to examine associations among diabetes distress, emotional wellbeing, and glycemic outcomes among adolescents with T1D. Additionally, interventions addressing anxiety and glycemic variability among adolescents with T1D should focus on diabetes distress as a means of ameliorating the link between anxiety and A1c. Effective components to address diabetes distress in adolescents include challenging negative thinking, engaging in positive problem-solving and assertive communication, and identifying social supports [37,46]. Similar strategies have yielded positive results for adults with T1D, particularly if they have higher baseline concerns with diabetes distress and/or emotion regulation [20]. These same strategies may yield positive effects for youth with T1D who have elevated anxiety and could impact automatic negative thinking [47,48], but further research is warranted.

Key Messages:

  • Anxiety has been significantly linked with A1c among adolescents with type 1 diabetes.

  • Current analyses demonstrate that anxiety exerts an indirect effect on A1c via automatic negative thinking and diabetes-related distress.

  • Diabetes distress mediates the anxiety-A1c link independent of automatic negative thinking and should be considered a treatment target.

Acknowledgement

The authors would like to thank the research teams at both Ann & Robert H. Lurie Children’s Hospital of Chicago and at Stanford University for their dedication to study implementation. Further, the authors kindly thank the families who devoted their time and effort to participating in this study. This study was funded by the National Institute of Diabetes and Digestive and Kidney Diseases (R01-090030).

Funding Statement:

This study was supported by funding from the United States National Institute of Diabetes and Digestive and Kidney Diseases (R01-090030).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author Disclosures: Dr. Hood reports grants from Dexcom Inc, personal fees from Cecelia Health, and personal fees from Cercacor that are not related to the submitted work. The other authors do not have any financial or ethical conflicts of interest.

Author Disclosures

The authors have no financial or ethical disclosures related to this work. Dr. Hood reports grants from Dexcom Inc, personal fees from Cecelia Health, and personal fees from Cercacor that are not related to the submitted work.

References

  • [1].Muris P, Schmidt H, Merckelbach H, Schouten E. Anxiety sensitivity in adolescents: factor structure and relationships to trait anxiety and symptoms of anxiety disorders and depression. Behav Res Ther 2001;39:89–100. 10.1016/S0005-7967(99)00179-5. [DOI] [PubMed] [Google Scholar]
  • [2].Bernstein CM, Stockwell MS, Gallagher MP, Rosenthal SL, Soren K. Mental health issues in adolescents and young adults with type 1 diabetes: Prevalence and impact on glycemic control. Clin Pediatr (Phila) 2013;52:10–5. 10.1177/0009922812459950. [DOI] [PubMed] [Google Scholar]
  • [3].Herzer M, Vesco A, Ingerski LM, Dolan LM, Hood KK. Explaining the family conflict-glycemic control link through psychological variables in adolescents with type 1 diabetes. J Behav Med 2011;34:268–74. 10.1007/s10865-010-9307-3. [DOI] [PubMed] [Google Scholar]
  • [4].Silverstein J, Klingensmith G, Copeland K, Plotnick L, Kaufman F, Laffel L, et al. Care of children and adolescents with type 1 diabetes: A statement of the American Diabetes Association. Diabetes Care 2005;28:186–212. 10.2337/diacare.28.L186. [DOI] [PubMed] [Google Scholar]
  • [5].Herzer M, Hood KK. Anxiety symptoms in adolescents with type 1 diabetes: Association with blood glucose monitoring and glycemic control. J Pediatr Psychol 2010;35:415–25. 10.1093/jpepsy/jsp063. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [6].Hilliard ME, Herzer M, Dolan LM, Hood KK. Psychological screening in adolescents with type 1 diabetes predicts outcomes one year later. Diabetes Res Clin Pract 2011;94:39–44. 10.1016/j.diabres.2011.05.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Fisher L, Hessler D, Polonsky W, Strycker L, Guzman S, Bowyer V, et al. Emotion regulation contributes to the development of diabetes distress among adults with type 1 diabetes. Patient Educ Couns 2018;101:124–31. 10.1016/j.pec.2017.06.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [8].Fisher L, Hessler D, Polonsky W, Strycker L, Bowyer V, Masharani U. Toward effective interventions to reduce diabetes distress among adults with type 1 diabetes: Enhancing Emotion regulation and cognitive skills. Patient Educ Couns 2019;102:1499–505. 10.1016/j.pec.2019.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [9].Watkins ER. Constructive and unconstructive repetitive thought. Psychol Bull 2008; 134:163–206. 10.1037/0033-2909.134.2.163. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Yapan S, Türkçapar MH, Boysan M. Rumination, automatic thoughts, dysfunctional attitudes, and thought suppression as transdiagnostic factors in depression and anxiety. Curr Psychol 2020. 10.1007/s12144-020-01086-4. [DOI] [Google Scholar]
  • [11].Puliafico AC, Kendall PC. Threat-related attentional bias in anxious youth: A review. Clin Child Fam Psychol Rev 2006;9:162–80. 10.1007/s10567-006-0009-x. [DOI] [PubMed] [Google Scholar]
  • [12].Sood ED, Kendall PC. Assessing anxious self-talk in youth: The negative affectivity self-statement questionnaire-anxiety scale. Cogn Ther Res 2007;31:603–18. 10.1007/s10608-006-9043-8. [DOI] [Google Scholar]
  • [13].Shapiro JB, Vesco AT, Weil LEG, Evans MA, Hood KK, Weissberg-Benchell J. Psychometric properties of the problem areas in diabetes: Teen and parent of teen versions. J Pediatr Psychol 2018;43:561–71. 10.1093/jpepsy/jsx146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [14].Fisher L, Mullan JT, Arean P, Glasgow RE, Hessler D, Masharani U. Diabetes distress but not clinical depression or depressive symptoms is associated with glycemic control in both cross-sectional and longitudinal analyses. Diabetes Care 2010;33:23–8. 10.2337/dc09-1238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Fisher L, Gonzalez JS, Polonsky WH. The confusing tale of depression and distress in patients with diabetes: A call for greater clarity and precision. Diabet Med 2014;31:764–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [16].Hagger V, Hendrieckx C, Sturt J, Skinner TC, Speight J. Diabetes distress among adolescents with type 1 diabetes: A systematic review. Curr Diab Rep 2016;16:1–14. 10.1007/s11892-015-0694-2. [DOI] [PubMed] [Google Scholar]
  • [17].Amsberg S, Anderbro T, Wredling R, Lisspers J, Lins P-E, Adamson U, et al. A cognitive behavior therapy-based intervention among poorly controlled adult type 1 diabetes patients—A randomized controlled trial. Patient Educ Couns 2009;77:72–80. 10.1016/j.pec.2009.01.015. [DOI] [PubMed] [Google Scholar]
  • [18].Sturt J, Dennick K, Hessler D, Hunter BM, Oliver J, Fisher L. Effective interventions for reducing diabetes distress: Systematic review and meta-analysis. Int Diabetes Nurs 2015;12:40–55. 10.1179/2057332415Y.0000000004. [DOI] [Google Scholar]
  • [19].Schmidt CB, Loon BJP van, Vergouwen ACM, Snoek FJ, Honig A. Systematic review and meta-analysis of psychological interventions in people with diabetes and elevated diabetes-distress. Diabet Med 2018;35:1157–72. 10.1111/dme.13709. [DOI] [PubMed] [Google Scholar]
  • [20].Fisher L, Hessler D, Polonsky WH, Masharani U, Guzman S, Bowyer V, et al. T1-REDEEM: A randomized controlled trial to reduce diabetes distress among adults with type 1 diabetes. Diabetes Care 2018;41:1862–9. 10.2337/dc18-0391. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].Briery BG, Rabian B. Psychosocial changes associated with participation in a pediatric summer camp. J Pediatr Psychol 1999;24:183–90. 10.1093/jpepsy/24.2.183. [DOI] [PubMed] [Google Scholar]
  • [22].Hains AA, Davies WH, Parton E, Silverman AH. Brief report: A cognitive behavioral intervention for distressed adolescents with type I diabetes. J Pediatr Psychol 2001;26:61–6. 10.1093/jpepsy/26.1.61. [DOI] [PubMed] [Google Scholar]
  • [23].Ellis DA, Frey MA, Naar-King S, Templin T, Cunningham PB, Cakan N. The effects of multisystemic therapy on diabetes stress among adolescents with chronically poorly controlled type 1 diabetes: Findings from a randomized, controlled trial. Pediatrics 2005;116:e826–32. 10.1542/peds.2005-0638. [DOI] [PubMed] [Google Scholar]
  • [24].Grey M, Boland EA, Davidson M, Yu C, Sullivan-Bolyai S, Tamborlane WV. Short-term effects of coping skills training as adjunct to intensive therapy in adolescents. Diabetes Care 1998;21:902–8. 10.2337/diacare.21.6.902. [DOI] [PubMed] [Google Scholar]
  • [25].Feldman MA, Anderson LM, Shapiro JB, Jedraszko AM, Evans M, Weil LEG, et al. Family-based interventions targeting improvements in health and family outcomes of children and adolescents with type 1 diabetes: A systematic review. Curr Diab Rep 2018;18:1–12. 10.1007/s11892-018-0981-9. [DOI] [PubMed] [Google Scholar]
  • [26].Kovacs M, Goldston D, Obrosky DS, Bonar LK. Psychiatric disorders in youths with IDDM: Rates and risk factors. Diabetes Care 1997;20:36–44. 10.2337/diacare.20.1.36. [DOI] [PubMed] [Google Scholar]
  • [27].Naar-King S, Idalski A, Ellis D, Frey M, Templin T, Cunningham PB, et al. Gender differences in adherence and metabolic control in urban youth with poorly controlled type 1 diabetes: The mediating role of mental health symptoms. J Pediatr Psychol 2006;31:793–802. 10.1093/jpepsy/jsj090. [DOI] [PubMed] [Google Scholar]
  • [28].Gonder-Frederick FA, Fisher CD, Ritterband LM, Cox DJ, Hou L, DasGupta AA, et al. Predictors of fear of hypoglycemia in adolescents with type 1 diabetes and their parents. Pediatr Diabetes 2006;7:215–22. 10.1111/j.1399-5448.2006.00182.x. [DOI] [PubMed] [Google Scholar]
  • [29].Majidi S, Driscoll KA, Raymond JK. Anxiety in children and adolescents with type 1 diabetes. Curr Diab Rep 2015;15:1–6. 10.1007/s11892-015-0619-0. [DOI] [PubMed] [Google Scholar]
  • [30].Hilliard ME, Wu YP, Rausch J, Dolan FM, Hood KK. Predictors of deteriorations in diabetes management and control in adolescents with type 1 diabetes. J Adolesc Health 2013;52:28–34. 10.1016/j.jadohealth.2012.05.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Kahkoska AR, Shay CM, Crandell J, Dabelea D, Imperatore G, Lawrence JM, et al. Association of race and ethnicity with glycemic control and hemoglobin A1c levels in youth with type 1 diabetes. JAMA Netw Open 2018;1:1–15. 10.1001/jamanetworkopen.2018.1851. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Kamps JL, Hempe JM, Chalew SA. Racial disparity in A1C independent of mean blood glucose in children with type 1 diabetes. Diabetes Care 2010;33:1025–7. 10.2337/dc09-1440. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].Petitti DB, Klingensmith GJ, Bell RA, Andrews JS, Dabelea D, Imperatore G, et al. Glycemic control in youth with diabetes: The SEARCH for diabetes in youth study. J Pediatr 2009;155:668–72. 10.1016/j.jpeds.2009.05.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Benkhadra K, Alahdab F, Tamhane SU, McCoy RG, Prokop LJ, Murad MH. Continuous subcutaneous insulin infusion versus multiple daily injections in individuals with type 1 diabetes: a systematic review and meta-analysis. Endocrine 2017;55:77–84. 10.1007/s12020-016-1039-x. [DOI] [PubMed] [Google Scholar]
  • [35].Vesco AT, Jedraszko AM, Garza KP, Weissberg-Benchell J. Continuous glucose monitoring associated with less diabetes-specific emotional distress and lower A1c among adolescents with type 1 diabetes. J Diabetes Sci Technol 2018;12:792–9. 10.1177/1932296818766381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [36].Hood KK, Iturralde E, Rausch J, Weissberg-Benchell J. Preventing diabetes distress in adolescents with type 1 diabetes: Results 1 year after participation in the STePS program. Diabetes Care 2018;41:1623–30. 10.2337/dc17-2556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [37].Weissberg-Benchell J, Rausch J, Iturralde E, Jedraszko A, Hood K. A randomized clinical trial aimed at preventing poor psychosocial and glycemic outcomes in teens with type 1 diabetes (T1D). Contemp Clin Trials 2016;49:78–84. 10.1016/j.cct.2016.05.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [38].Spielberger CD. State-trait anxiety inventory for adults. Palo Alto, CA: Mind Garden; 1983. [Google Scholar]
  • [39].Hollon SD, Kendall PC. Cognitive self-statements in depression: Development of an automatic thoughts questionnaire. Cogn Ther Res 1980;4:383–95. 10.1007/BF01178214. [DOI] [Google Scholar]
  • [40].Hayes AF. Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York: Guilford Press; 2013. [Google Scholar]
  • [41].Rosseel Y. Lavaan: An R package for structural equation modeling and more Version 0.5-12 (BETA). J Stat Softw 2012;48:1–36. [Google Scholar]
  • [42].R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2020. [Google Scholar]
  • [43].Bastelaar KMPV, Pouwer F, Geelhoed-Duijvestijn PHLM, Tack CJ, Bazelmans E, Beekman AT, et al. Diabetes-specific emotional distress mediates the association between depressive symptoms and glycaemic control in type 1 and type 2 diabetes. Diabet Med 2010;27:798–803. 10.1111/j.1464-5491.2010.03025.x. [DOI] [PubMed] [Google Scholar]
  • [44].Hessler DM, Fisher L, Polonsky WH, Masharani U, Strycker LA, Peters AL, et al. Diabetes distress is linked with worsening diabetes management over time in adults with type 1 diabetes. Diabet Med 2017;34:1228–34. 10.1111/dme.13381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [45].Fonda SJ, McMahon GT, Gomes HE, Hickson S, Conlin PR. Changes in diabetes distress related to participation in an internet-based diabetes care management program and glycemic control. J Diabetes Sci Technol 2009;3:117–24. 10.1177/193229680900300113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [46].Weissberg-Benchell J, Shapiro JB, Bryant FB, Hood KK. Supporting Teen Problem-Solving (STePS) 3 Year Outcomes: Preventing Diabetes-Specific Emotional Distress and Depressive Symptoms in Adolescents With Type 1 Diabetes. J Consult Clin Psychol 2020;88:1019–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [47].Albano AM, Kendall PC. Cognitive behavioural therapy for children and adolescents with anxiety disorders: clinical research advances. Int Rev Psychiatry Abingdon Engl 2002;14:129–34. [Google Scholar]
  • [48].Kendall PC, Peterman JS. CBT for Adolescents With Anxiety: Mature Yet Still Developing. Am J Psychiatry 2015;172:519–30. [DOI] [PubMed] [Google Scholar]

RESOURCES