Observed Collaborative and Intrusive Parenting Behaviors Associated with Psychosocial Outcomes of Adolescents with Type 1 Diabetes and their Maternal Caregivers

Abstract

Background:

Maternal caregiver involvement is strongly associated with psychosocial and glycemic outcomes amongst adolescents with type 1 diabetes (T1D); however, previous studies have lacked detailed, objective examinations of caregiver involvement. We examined the relationship between observed parenting behaviors and psychosocial and glycemic outcomes amongst youth with T1D.

Methods:

Data collected from adolescents with T1D (age 11–17) and their female caregivers as a part of a randomized controlled trial were analyzed. These included structured, observation-based scores of adolescent-caregiver dyads engaged in videotaped interactions and selected psychosocial and glycemic outcome measures.

Results:

In adjusted analyses, higher levels of intrusive parenting behaviors during observed interactions were associated with higher diabetes distress in adolescents, but no difference in HbA1c. Associations between intrusive parenting behaviors and psychosocial outcomes were stronger for females compared to males for both diabetes distress and quality of life. Similarly, associations between collaborative parenting behaviors and quality of life were stronger for female adolescents than males. No associations were observed between collaborative parenting behaviors and glycemic outcomes. Consistent with previous work, we noted higher levels of adolescent-reported family conflict were associated with lower adolescent quality of life and higher diabetes distress with no significant difference between male and female adolescents.

Conclusion:

These findings indicate that high levels of intrusive parenting behaviors, such as lecturing or over-controlling behaviors, are associated with lower levels of adolescent well-being, particularly among adolescent girls. This work suggests that interventions to reduce intrusive parenting by maternal caregivers could result in improved psychosocial outcomes for adolescents with T1D.

1. Introduction:

Adolescents with type 1 diabetes (T1D) and their caregivers experience a unique set of challenges as youth move toward increasing independence of their diabetes self-management.¹ This shift in the burden of disease management is associated with increased risk for psychosocial problems in adolescents and their caregivers, with clinically significant diabetes distress reported by 30–45% of adolescents, and clinically significant depressive symptoms evident in up to 61% of parents. ^2–4 These adverse psychosocial outcomes are also associated with unfavorable glycemic outcomes, an area of particular concern, as this age group has the highest average HbA1c of any population with T1D.^5,6 Higher HbA1c increases risk for short-term complications, including diabetic ketoacidosis, along with long-term risk of microvascular and macrovascular complications.⁷ While large longitudinal cohort studies have identified gender, age of diabetes onset and diabetes technology use as predictors of glycemic trajectories, these studies did not evaluate the role of potentially modifiable family-level factors on these outcomes.^8–10

Caregiver involvement is strongly associated with psychosocial and glycemic outcomes among adolescents with T1D.¹¹ Specifically, the quality of caregiver involvement is a strong predictor of adolescent adjustment, and positive parenting behaviors have been associated with better quality of life, fewer depressive symptoms, better self-management, and decreases in HbA1c in adolescents with T1D.^12–16 In contrast, diabetes-related family conflict is strongly associated with poorer outcomes in youth with T1D, and intrusive or overinvolved parenting in particular predict negative psychosocial outcomes in adolescents with T1D.^16,17

The Transactional Stress and Coping Model, which provides a framework to conceptualize caregiver-child relationships, has previously been adapted and used to analyze factors influencing the psychosocial outcomes of adolescents with T1D.¹⁸ Consistent with previous observations, this model suggests that caregiver distress influences the quality of their interactions with their child, which subsequently impacts adolescent distress and ultimately glycemic outcomes (Figure 1). Interactions between adolescents and maternal caregivers may be particularly relevant to assess, since mothers of youth with T1D report higher rates of distress than fathers, and maternal distress has been strongly associated with outcomes in youth with T1D.^19–21 Despite serving as the key link between caregiver and adolescent psychosocial outcomes, there is little objective data documenting the specific behaviors underlying these constructs. For example, one of the most widely used measures of diabetes-related family conflict asks parents and adolescents to report on how frequently they argue about tasks related to diabetes management over the past month, but this does not provide information about modifiable behaviors.^14,16,17

Figure 1:

Conceptual Framework for Analysis (informed by the Transactional Stress and Coping Model)

Furthermore, despite evidence in population level retrospective data, previous studies evaluating the impact of family conflict on parent and adolescent outcomes have not evaluated in detail the potential role of adolescent gender as a moderator of outcomes. Prior studies suggest that maternal parenting style in general and controlling behaviors in particular may have a stronger impact on psychosocial outcomes in female adolescents as compared to males, but these studies relied on self-report rather than direct observation of these behaviors.^22–26 Developing a better understanding of the specific types of parenting behaviors are associated with adolescent outcomes and how these might differ based on factors such as adolescent gender is an important next step toward addressing conflict in these critical relationships.

The Iowa Family Interaction Rating Scales (IFIRS) is a validated observational coding system used to evaluate and quantify parenting behaviors observed during caregiver-child interactions.²⁷ Building on previous work which has shown specific IFIRS-coded behaviors to be associated with HbA1c and psychosocial outcomes in adolescents, we hypothesized that composite measures of IFIRS-scored observed parenting behaviors would be associated with adolescent- and maternal caregiver-reported measures of diabetes distress and family functioning.

2. Methods:

2.1. Participants & Procedures:

The current study is an analysis of baseline data from participants who enrolled in a randomized trial of a behavioral intervention and completed a videotaped caregiver-child interaction. Families were eligible for the trial if the adolescent (age 11–17) had a diagnosis of T1D for at least 12 months, and maternal caregivers reported at least moderate distress or depressive symptoms. There was no eligibility criterion related to HbA1c. All caregivers were either mothers or the primary female caregiver of study participants. Full details of inclusion/exclusion criteria and study protocol have been previously reported.²⁸ All study procedures were reviewed and approved by the Vanderbilt University Institutional Review Board (IRB#182217) prior to study initiation. After providing informed consent/assent, both caregivers and adolescents completed questionnaires consistent with our theoretical model (Figure 1) and each dyad completed a 15-minute videotaped conversation about diabetes-related stress. The topic chosen for this interview was selected based on the mother’s highest-rated diabetes related stressor, with prompts to guide the video-taped discussion (e.g., What happened the last time [we forgot diabetes supplies]?). While the first 49 dyads completed these conversations in a private space adjacent to the diabetes clinic, due to disruption caused by the COVID-19 pandemic, those who completed data collection at later times (n=63) completed their conversations via a secure, online videoconferencing tool. This was accomplished by having a member of the research team videotape a videoconference with the dyad, who were instructed to face each other and interact as they normally would. Despite efforts to complete videotaped interactions with each of the 151 adolescent-caregiver dyads enrolled in the trial, 39 dyads we unable to complete this during the study baseline period.

2.2. Observed Parenting Behaviors:

The IFIRS is an global observational coding system designed to measure family functioning which has been validated and is well established for use in pediatric populations.^29,30 In this study, each videotaped interaction was reviewed by two trained IFIRS coders who scored selected maternal caregiver behaviors on a scale of 1 (absent) to 9 (high level) based on the behavior’s frequency, context, affect, intensity, and proportion. Inter-rater reliability for individual codes (intraclass correlation coefficients) ranged from .58 −.76. The mean of the two coders’ scores were used in analyses, with score discrepancies >2 reviewed and adjudicated by a third reviewer (a third rater was used for 9 videos). Once reviews were completed, composite codes were calculated based on previous methods.^31,32 The Collaborative Parenting composite includes 3 codes: 1) Communication, the extent to which the mother uses appropriate explanations and gives appropriate feedback (e.g., “I know that you’re busy, but I think you could check your blood sugar more”); 2) Positive Reinforcement, the extent to which the mother responds to her child’s desired behavior with praise or rewards (e.g., “I’m really proud of how you’ve dealt with all of this”); and 3) Child Centered, the degree to which the mother displays an awareness of the child’s needs, moods, interests, and capabilities, and is “in sync” with the child’s interactions. The Intrusive Parenting composite also included 3 codes: 1) Parental Influence, parental attempts to regulate, control or influence the child’s behavior (e.g., “You need to check your blood sugar before you drive to the game tonight”); 2) Intrusiveness, overcontrolling behaviors that are parent-centered and emphasize task completion rather than promoting the child’s autonomy (e.g., “I don’t care if you don’t want to talk about your high blood sugars, we have to”); and 3) Lecturing/Moralizing, the extent to which the parent tells the child how to think in a way that assumes superior wisdom and provides little opportunity for the child to think independently (e.g., “You should know better than to leave home without your supplies”). Scores for each composite scale range from 3 to 27, with higher scores indicating increased frequency/intensity of the parenting behaviors. Inter-rater reliability (κ) for the composite scales was good; the intraclass correlation was 0.76 for Collaborative Parenting and 0.80 for Intrusive Parenting.

2.3. Psychosocial Measures:

2.3.1. Adolescent Diabetes Distress (PAID-T):

Adolescent distress was measured using the Problem Areas in Diabetes – Teen Version (PAID-T), a measure of diabetes distress adapted for use in teens.³³ The PAID-T questionnaire consists of 26 items with scores ranging from 26 to 156 with higher numbers indicating higher diabetes distress.³⁴ Cronbach’s alpha was 0.96.

2.3.2. Pediatric Quality of Life (PedsQL):

Adolescents’ diabetes-related quality of life was measured using the PedsQL 3.2 diabetes module, a 28-item questionnaire validated for use in children and adolescents with T1D.³⁵ Scores range from 0 to 100, with higher scores indicating increased quality of life. In the current sample, Cronbach’s alpha was 0.91.

2.3.3. Diabetes-related Family Conflict (DFCS):

Family conflict related to diabetes was assessed using the Revised DFCS.¹⁷ This instrument is a 19-item questionnaire which has been adapted for use both for adolescents and caregivers. The 19 items are scored 0 to 3 with higher scores indicating higher levels of conflict. Cronbach’s alpha for child reported DFCS was 0.93 and 0.84 for maternal caregiver reported DFCS.

2.4. Glycemic Outcomes:

Hemoglobin A1c (HbA1c) is a measure of blood glucose levels over the prior 8 to 12 weeks. The point-of-care value from the clinic visit on the day of enrollment was extracted from participants’ medical records. For those participants who enrolled at a time other than at a clinic visit, their most recent HbA1c was obtained (after consent) from their medical record.

2.5. Statistical Methods:

All data entry was completed via REDCap and statistical analyses were completed using STATA (v. 17.0, College Station, TX).³⁶ First, we calculated descriptive statistics, then examined associations using Pearson correlations between our composite parenting behavior scores and adolescent and caregiver reports of diabetes-related family conflict. Then, guided by our theoretical model (Figure 1), we used linear regression to examine the associations between parenting behavior composite scores and family conflict scales and glycemic and psychosocial outcomes. In each of these analyses, we constructed linear regression models adjusting for adolescent age, self-identified gender, diabetes duration and self-reported race/ethnicity. These covariates were included as parental involvement and parenting styles evolve as children age as does their perception of family conflict. Further, social norms and cultural expectations of child-rearing vary with racial background which also is associated with diabetes outcomes.⁶ Finally, previous work has also shown duration of diabetes diagnosis to be associated glycemic and psychosocial outcomes.^8,9 Thus, we included these variables to mitigate the confounding effects of these factors. Finally, given significant differences in observed outcomes between male and female participants, as well as previous work suggesting differences in family conflict between male and female adolescents, we used linear regression to examine the moderation by child gender on the relationship between observed and self-reported parenting behaviors on our primary outcomes of interest in a post-hoc, exploratory analysis.²³

3. Results:

3.1. Sample demographics and preliminary analyses:

Adolescent demographics and subjective and objective measures of caregiver behavior are summarized in Table 1. Adolescent participants self-identified as 55% female, 83% non-Hispanic White and mean age was 14.1±2 years old. For diabetes management, 63% used insulin pumps while 76% used continuous glucose monitors, and mean HbA1c of study participants was 75 mmol/mol ± 21 (9.0% ± 1.9%). Male and female participants did not significantly differ in terms of race or diabetes technology usage, but male participants were nearly 1 year older on average (14.6yrs vs. 13.6yrs) and had a higher average HbA1c (78 mmol/mol [9.3%] vs. 73 mmol/mol [8.8%]). Demographics of dyads completing videotaped interactions, were similar to the parent study cohort (Supplemental Table 1). In terms of measures of family interactions, DFCS scores were not significantly different on average between child and caregiver report (29.1 vs. 30.0). In addition, while adolescent-reported DFCS did not differ between male and female adolescents (28.5 vs. 29.6, p = 0.54), caregiver-reported DFCS was significantly higher among caregivers of male adolescents compared to those with female adolescents (31.5 vs. 28.8, p = 0.03). Amongst the observational measures of parenting behavior, the Collaborative composite did not differ significantly between male and female adolescent participants (13.9 vs. 14.8, p = 0.27), but the intrusive behavior composite was significantly higher among maternal caregivers of male adolescents compared to female adolescents (11.4 vs. 9.5, p = 0.03). Descriptive statistics of individual observation based IFIRS code scores are presented in Supplemental Table 2.

Table 1:

Participant Characteristics and Behavior Measures

	Female Adolescents (N = 62)	Male Adolescents (N = 50)	All Adolescents (N=112)
Adolescent Demographic Characteristics
Age (years) – mean (SD)	13.6 (2.0)	14.6 (1.9)	14.1 (2.0)
Self-Reported Race – n(%)
Non-Hispanic White	53 (86%)	40 (80%)	93 (83%)
Black	5 (8%)	7 (14%)	12 (11%)
Other	4 (6%)	3 (6%)	7 (6%)
Using Insulin pump – n(%)	38 (61%)	32 (64%)	70 (63%)
Using Continuous Glucose Monitor – n(%)	47 (76%)	38 (76%)	85 (76%)
Years since diagnosis – mean (SD)	8.0 (3.2)	8.0 (3.8)	8.0 (3.4)
HbA1c – mean (SD)	73 (22) [8.8% (2.0)]	78 (20) [9.3% (1.8)]	75 (21) [9.0% (1.9)]
Caregiver Relationship to Adolescent
Mother	59 (95%)	50 (100%)	109 (97%)
Other female family member	3 (5%)	0	3 (3%)
Subjective Caregiver Behavior Measures– mean (SD)
DFCS - Child	29.6 (8.7)	28.5 (8.9)	29.1 (8.8)
DFCS - Caregiver	28.8 (6.0)	31.5 (6.1)	30.0 (6.1)
Observed Caregiver Behavior (IFIRS) by code* – mean (SD)
Collaborative (CO+PO+CC)	14.8 (4.3)	13.9 (3.9)	14.4 (4.1)
Intrusive (PI+NT+LM)	9.5 (4.2)	11.4 (4.8)	10.3 (4.5)

^*Raw Observation Scores available in Supplemental Table 1

3.2. Associations between self-reported and observed measures of female caregiver involvement:

In preliminary analyses (Table 2), we evaluated the correlations between self-report and observation-based measures of caregiver involvement including DFCS scores, and Collaborative and Intrusive composite scores. We found significant positive associations between adolescent-and caregiver-reported DFCS (r = 0.32), and a significant negative association between the observed parenting behavior composite scores (r = −0.39). Both adolescents’ and caregivers’ reports of family conflict were significantly associated with higher levels of observed Intrusive behavior (r= 0.19, 0.40, respectively) and with lower levels of Collaborative behavior (r= −0.24, −0.25, respectively). We also observed significant negative associations between adolescents’ report of family conflict and quality of life (r = −0.42) while higher levels of caregiver-reported conflict were associated with higher HbA1c (r = 0.41).

Table 2:

Associations Between Measures of Caregiver Involvement and Adolescent Glycemic and Psychosocial Outcomes.

	Variables	1.	2.	3.	4.	5.	6.	7.	8.
1.	DFCS - Child	---
2.	DFCS - Caregiver	0.32*	---
3.	Collaborative Composite (CO+PO+CC)	−0.24*	−0.25*	---
4.	Intrusive Composite (PI+NT+LM)	0.19*	0.40*	−0.39*	---
5.	Diabetes Distress (PAID-T)	0.32*	0.30*	−0.06	0.19	---
6.	Quality of Life (PedsQL)	−0.42*	−0.13	0.11	−0.14	−0.80*	---
7.	HbA1c (mmol/mol)	0.09	0.41*	−0.10	0.15	0.17	−0.19	---
8.	Adolescent Age	0.01	0.17	−0.03	0.13	0.06	0.04	0.13	---
9	Diabetes Duration	−0.02	−0.09	0.13	−0.01	0.12	−0.07	−0.10	0.17

DFCS = Diabetes Family Conflict Scale; CO = Communication; PO = Positive Reinforcement; CC = Child Centered; PI = Parental Influence; NT = Intrusiveness; LM = Lecture/Moralizing

^*Indicates association is significant at the p ≤ 0.05 level.

3.3. Observed parenting measures and psychosocial and glycemic outcomes:

To further evaluate associations predicted in our conceptual model (Figure 1), we used linear regression to evaluate the relationships between observation-based measures of maternal caregiver involvement and psychosocial and glycemic outcomes (Table 3). In our adjusted primary analysis, we did not observe associations between the collaborative parenting composite (Table 3) and either of our measures of psychosocial outcomes (PAID-T and PedsQL). However, our analysis of the Intrusive composite score did show a significant association in our adjusted model with higher levels of the Intrusive parenting composite score significantly associated with higher diabetes distress (β = 0.21; p = 0.03), but not lower quality of life. HbA1c was not associated with either composite score in our adjusted model.

Table 3:

Regression Analysis Predicting Adolescents’ Psychosocial Outcomes from Demographic Factors and Collaborative Parenting Behaviors

Predictor	Collaborative Composite					Intrusive Composite
	PAID-T		PedsQL		PAID-T		PedsQL
	β	ΔR2	β	ΔR2	β	ΔR2	β	ΔR2
Covariates Only		.18**		.19**		.18**		.19**
Age	.14		−.05		.14		−.05
Duration of Diabetes	.12		−.11		.12		−.11
Race/Ethnicity	.18		−.24*		.18		−.24*
Gender	.39**		−.39**		.39**		−.39**
Adjusted Model		.00		.01		.04*		.03
Age	.14		−.04		.11		−.03
Duration of Diabetes	.13		−.12		.13		−.12
Race/Ethnicity	.17		−.22*		.13		−.21*
Gender	.39**		−.40**		.41**		−.42**
Parenting Composite	−.05		.12		.21*		−.17
Gender Moderation		.01		.04*		.08**		.04*
Age	.13		−.02		.11		−.02
Duration of Diabetes	.12		−.12		.11		−.10
Race/Ethnicity	.17		−.23*		.11		−.18
Gender	.38**		−.39**		.41**		−.42**
Parenting Composite	.08		−.13		−.07		.04
Gender × Composite	.17		.31*		.40**		−.29*

^*Indicates associations at the p ≤ 0.05 level

^**Indicates associations at the p ≤ 0.01 level

3.4. Gender-based effect modification between observed parenting measure and psychosocial outcomes:

We explored the potential of gender-based effect modification by adding a gender × parenting composite term to the regression models. In the final model (Table 3), we observed a significant moderating effect of gender on the association between the Collaborative parenting composite and quality of life (β = 0.27; p = 0.04). We also observed a significant moderating effect of gender on the association between the Intrusive composite and quality of life (β = −0.34; p = 0.008) and the association between the Intrusive composite and diabetes distress (β = −0.46; p <0.001). As seen in Figure 2A–C, the adjusted associations between parenting behaviors and psychosocial outcomes were significantly stronger for females than for males. As with our primary analysis, the relationship between HbA1c and both of the observation-based composites did not show significant moderation by gender.

Figure 2:

Adjusted associations between parenting behavior measures and outcomes by adolescent gender.

3.5. Adolescent and Caregiver-reported family conflict and psychosocial and glycemic outcomes:

In our analysis of the questionnaire measures of family conflict, we conducted regression analyses to examine associations between adolescents’ and maternal caregivers’ report of family conflict and adolescents’ psychosocial and glycemic outcomes (Table 4). We found that adolescent-reported family conflict (DFCS-Child) was significantly associated with both adolescents’ report of diabetes distress (β = 0.31; p = 0.001) and quality of life (β = −0.42; p <0.001), but not with HbA1c in our adjusted model (Table 4). Similarly, we found higher levels of caregiver reported family conflict (DFCS-Caregiver) were significantly associated with greater adolescent diabetes distress (β = 0.37; p <0.001), lower quality of life (β = −0.22; p = 0.03) and higher HbA1c (β = 0.37; p <0.001).

Table 4:

Regression Analysis Predicting Adolescents’ Psychosocial Outcomes from Demographic Factors and Diabetes Family Conflict Scores

Predictor	DFCS - Child				DFCS - Caregiver
	PAID-T		PedsQL		PAID-T		PedsQL
	β	ΔR2	β	ΔR2	β	ΔR2	β	ΔR2
Covariates Only		.18**		.19**		.18**		.19**
Age	.14		−.05		.14		−.05
Duration of Diabetes	.12		−.11		.12		−.11
Race/Ethnicity	.18		−.24*		.18		−.24*
Gender	.39**		−.39**		.39**		−.39**
Adjusted Model		.09**		.18**		.12**		.04*
Age	.13		−.02		.07		.01
Duration of Diabetes	.11		−.10		.20*		−.18
Race/Ethnicity	.13		−.19*		.15		−.26**
Gender	.37**		−.36**		.41**		−.41**
DFCS	.31**		−.42**		.37**		−.22*
Gender Moderation		.02		.01		.05*		.02
Age	.12		−.01		.07		.01
Duration of Diabetes	.12		−.11		.20*		−.18
Race/Ethnicity	.13		−.19*		.18		−.27**
Gender	.37**		−.37**		.39**		−.39**
DFCS	.17		−.30*		.11		−.07
Gender × DFCS	.19		−.16		.33*		−.19

^*Indicates associations at the p ≤ 0.05 level

^**Indicates associations at the p ≤ 0.01 level

3.6. Gender-based effect modification between adolescent and caregiver-reported family conflict measures and psychosocial outcomes:

Similar to the analysis of observed parenting measures, we conducted moderation analysis by gender to further explore differences in these relationships between male and female adolescents (Table 4).

In our analyses of gender-based moderation in the associations between caregiver-reported family conflict and psychosocial and glycemic outcomes, we observed a statistically significant effect of gender on the association between caregiver-reported family conflict and adolescent distress (β = 0.33; p = 0.015). As seen in Figure 2D, caregiver’s report of family conflict being was more strongly associated with adolescent diabetes distress among females than females. Moderation by gender was not observed between caregiver’s report of family conflict and either adolescent quality of life or HbA1c.

4. Discussion:

The current study is one of the first to provide a multifaceted view of interactions between adolescents with T1D and their maternal caregivers by including an objective measure of family interactions along with caregiver and adolescent reports of diabetes-related family conflict. Our structured, observational approach to measuring these interactions provides important insight into the specific and potentially modifiable parenting behaviors which may influence adolescent psychosocial and glycemic outcomes. Our primary analysis demonstrated a relationship between intrusiveness in parenting interactions and markers of adolescent well-being, including significantly higher levels of diabetes distress. While we did not find a significant relationship between observed parenting behaviors and glycemic outcomes, we did find that parenting behaviors were associated with both adolescent and caregiver-reports of family conflict, which were also correlated with HbA1c. These data suggest that adolescent psychosocial outcomes may be influenced by caregiver interactions, but that any relationship between these interpersonal interactions and glycemic outcomes is likely blunted by multiple other physiologic and psychosocial factors all of which ultimately determine overall glycemic control.

In particular, the parenting behaviors included in the Intrusive composite (parental influence, intrusiveness, and lecture/moralizing) are behaviors which move the focus of the discussion from the child’s strengths to their perceived deficiencies and prevent adolescents from exercising their autonomy in their own diabetes care. It is important to note that these were not hostile or aggressive parenting behaviors, but rather, statements about the importance of diabetes management made by concerned, well-meaning maternal caregivers. In line with the Transactional Stress and Coping model, it is likely that these intrusive parenting behaviors increase when adolescents’ diabetes management is not going well. The association between these parenting behaviors with poorer psychosocial outcomes suggests that adolescents (and particularly girls) are sensitive to these behaviors, possibly because they undermine adolescents’ attempts at gaining independence in self-care – a key developmental task in this age group.²³

The Collaborative parenting composite, which includes observed communication, positive reinforcement, and child-centered behaviors, was not associated either with psychosocial or glycemic outcomes in our primary analyses but was associated with lower levels of diabetes distress in girls. Generally, these parenting behaviors communicate explicitly and implicitly that the caregiver recognizes the perspective of the child (communication), validate the strengths of their ideas and actions (positive reinforcement), and display an awareness of the child’s feelings (child-centered) despite discussing difficult topics. Previous data suggest that collaborative parenting is associated with improved psychosocial and glycemic outcomes in this population.^37,38 The lack of a significant association in our data may suggest either that our composite score may not adequately capture all of the beneficial aspects of collaborative caregiver behaviors or that the degree of association between these supportive behaviors and outcomes are relatively modest compared to the degree of their association with observed Intrusive behaviors, making the sample underpowered to observe these more modest associations with the collaborative composite.

Consistent with previous observations, we found significant associations between both maternal caregiver and adolescent reports of diabetes conflict and measures of adolescent well-being (Table 4). Notably, we observed an association between HbA1c and caregiver-reported family conflict, but not adolescents’ report of conflict. This suggests that caregivers may be more highly focused on glycemic outcomes both because of the impact on long-term health and because of the easily observed nature of HbA1c. In contrast to their maternal caregivers, and in line with our theoretical model, adolescents’ experience of family conflict appears to have more direct relationship to their psychosocial well-being, regardless of their glycemic outcomes, potentially explaining the stronger associations between these measures.

Aside from the significant relationships with psychosocial and glycemic outcomes, the association of the observation-based parenting behavior composite measures with questionnaire measures of family conflict supports the use of these self-report measures in assessment of these relationships. It is important to note, however, that maternal report of family conflict was more strongly correlated with intrusive behaviors in our sample than adolescents’ reports of family conflict. Therefore, obtaining measures of family conflict from both parents and adolescents may be useful, as they may be reflective of different aspects of these relationships. Importantly, the observational data in this study provide significantly more detail into the types of parenting behaviors that are potentially influencing diabetes outcomes in this population than are measured by survey-based instruments.

Our analyses exploring gender as a modifier of the relationship between parenting behaviors and psychosocial outcomes suggests that adolescent females with T1D may be more impacted by intrusive parenting behaviors than their male counterparts. While our analysis is exploratory in nature, previous work has shown that maternal parenting style has stronger associations with psychosocial outcomes (depressive symptoms and self-efficacy) in adolescents girls than boys with T1D.^23,26 Thus, although we observed higher levels of intrusive parenting behaviors among maternal caregivers of adolescent males vs. females, girls may be more responsive to parenting style than boys.³⁹ Given the post-hoc nature of these analyses, these data should be interpreted with caution, but do suggest that child gender should be more carefully explored in future studies particularly among high-risk adolescents with T1D.

This study had several strengths, including a large sample size of participants with observed family interactions and multiple measures of diabetes-related outcomes. Furthermore, the selection criteria of eligible dyads – requiring some level of psychosocial distress in mothers – ensured that these data are relevant to those families that stand to benefit most from this knowledge. This study also has several limitations, including the change in study procedures due to the COVID-19 pandemic. While efforts were taken to mitigate the impact of this change and the populations completing in-person vs. virtual recordings were relatively similar, this cannot be excluded as a potential source of bias. Additionally, while our focus on distressed maternal caregivers and adolescents provides important information about this high-risk population, it also does limit the generalizability of the data to less distressed families and those with male primary caregivers.⁴⁰

Despite these limitations, this study provides important data to describe specific and potentially modifiable caregiver behaviors to guide future research and clinical care to decrease family conflict and improve adolescent outcomes. Specifically, these data suggest that while increasing collaborative interactions may be beneficial, minimizing intrusive parenting behaviors may have a more significant impact on outcomes, particularly among adolescent females with T1D. Future work should work to further evaluate how adolescent boys and girls may be differentially impacted by parenting behaviors and work to expand the generalizability of these findings by including male caregivers and dyads with less psychological distress.

Novelty Statement:

• Observational assessment of interactions between adolescents and their caregivers provides an additional level of detail to evaluate these relationships.

• Higher levels of observed intrusive parenting behaviors were associated with higher levels of adolescent diabetes distress.

• The association between intrusive behavior and psychosocial outcomes was significantly stronger among female adolescents compared to males with T1D.

• Future work should focus on further exploring the role of adolescent gender in diabetes family conflicts and in developing and testing interventions to reduce intrusive parenting behaviors to improve outcomes amongst adolescents with type 1 diabetes.