Communal Coping with Type 2 Diabetes: A Five-Year Measurement Burst Study

Abstract

Objective:

Communal coping is an interpersonal coping style that has been linked to positive psychosocial and health outcomes. The study goals were (1) to investigate changes in communal coping among persons with diabetes (PWD) over five years and (2) to assess how links of communal coping to outcomes change over that time.

Methods:

A measurement burst design was used. Couples in which one person had type 2 diabetes (64% White, 36% Black) completed a 14-day diary shortly after diagnosis (M = 1.88 years) (2012–2017) and again five years later. Mean levels of communal coping (shared appraisal, collaboration) among PWD were compared across the two bursts to assess changes in communal coping. Multilevel modeling was used to assess links of between- and within-person communal coping to psychosocial (mood, coping, positive support, negative interactions) and diabetes (glucose checking, glucose level, dietary adherence) outcomes. Interactions with time were included to determine how links of communal coping to outcomes changed over time.

Results:

Communal coping decreased across the five-years among the 99 PWD. Consistent with past research, within- and between-person communal coping were linked to positive psychosocial outcomes and improved diet. Overall, between-person communal coping was more strongly linked to positive outcomes at Time 2 than Time 1. The opposite pattern was observed at the within-person level, but it was less consistent for diabetes outcomes, and several exceptions emerged.

Conclusions:

Person-level communal coping becomes more important over time. Interventions aimed at sustaining communal coping may facilitate better health among people with type 2 diabetes.

Type 2 diabetes is one of the most prevalent chronic diseases among adults, and its rate has increased dramatically over the past 20 years (Tönnies et al., 2023). Type 2 diabetes affects 11.6% of the U.S. population, is the 8^th leading cause of death (Centers for Disease Control and Prevention, 2024), and is associated with a number of long-term health complications, such as neuropathy, retinopathy, heart disease, and kidney problems (Haw et al., 2021). Preventing these complications involves adhering to a labor-intensive self-care regimen, involving changes in diet, physical exercise, monitoring blood glucose, and taking one or more medications to control blood glucose levels and associated conditions (e.g., blood pressure, lipid management). Individuals often find balancing the demands of diabetes management with other life responsibilities to be a challenge, requiring considerable organization, focus, and self-discipline. Thus, not surprisingly, diabetes is associated with impaired quality of life, including psychological distress and interpersonal difficulties (Hermanns et al., 2018).

Although much of the disease management research in type 2 diabetes focuses on the person with diabetes (PWD) and their behaviors (e.g., self-efficacy, Gonzalez et al., 2016), disease management takes place in an interpersonal context that extends beyond the patient. There is a large literature that shows social support is related to the self-management of chronic illness, including a meta-analytic review that links social support to better diabetes self-management (Song et al., 2017).

Among the sources of support, research suggests that spouses and romantic partners play a prominent role and can be important resources upon which PWD can draw. Spouses are often involved in a PWD’s illness management (Franks et al., 2012), but this involvement can be helpful or harmful (Mayberry et al., 2019). In a daily diary study of couples in which one person had type 2 diabetes, spouse support (emotional and instrumental support) was related to greater daily activity, whereas spouse controlling behavior (reminding PWD about diabetes care, watching or criticizing self-care efforts) was unrelated to physical activity (Khan et al., 2013). Another daily diary study showed that partner emotional support was linked to better mood, more exercise, and greater dietary adherence on a daily basis, whereas partner controlling behavior (criticizing, arguing, or nagging the PWD about diabetes care) was linked to worse mood and unrelated to self-management behaviors (Helgeson et al., 2016).

We have argued that one beneficial way that romantic partners can be involved in self-management that extends beyond social support is via communal coping. Communal coping involves viewing diabetes as a shared problem and working together to manage the disease (Helgeson et al., 2018; Lyons et al., 1998). When couples view a problem as shared, it makes it easier for the patient to request support, for partners to provide support, and for the two to communicate about the illness. When couples collaborate or work together on the problem, there may be more optimal outcomes with less energy expended.

Indeed, there is substantial research that shows communal coping with chronic illness is related to good patient outcomes. In a qualitative study of couples in which one person had type 2 diabetes, focus groups revealed that patients were most likely to exercise when the couple adopted a team approach to diabetes, there was a shared sense of responsibility, and they were ‘in this together’—that is, the couple engaged in communal coping (Beverly & Wray, 2010). We-language (i.e., use of first-person plural pronouns) in the context of illness discussions is considered to reflect the shared appraisal aspect of communal coping and has been linked to better quality of life and reduced physical symptoms among patients with heart failure (Rohrbaugh et al., 2008) and greater abstinence following a smoking cessation intervention (Rohrbaugh et al., 2012). Indeed, a meta-analytic review of pronoun usage showed that we-language was linked to better patient physical health and the adoption of good health behaviors (Karan et al., 2019). There is also evidence for benefits of collaboration. In a 14-day daily diary study of men with prostate cancer, daily collaboration with one’s partner on diabetes stressors was related to more positive emotions (Berg et al., 2008).

We have studied communal coping in the context of couples in which one person was relatively recently diagnosed with type 2 diabetes—our goal being to understand whether couples who initially respond to the disease with communal coping would have better outcomes. We have found substantial evidence that this is the case. First, self-report measures of communal coping have been linked to better diabetes self-care and lower patient distress (Helgeson, Jakubiak, et al., 2016). Second, observed communal coping from videotaped couple conversations have been linked to better relationship functioning, reduced psychological distress (Helgeson et al., 2019), and greater progress in resolving diabetes problems (Van Vleet et al., 2018). Third, in a 14-day diary examination of a subset of these couples (n = 123), daily reports of communal coping were related to better same-day mood and self-care behavior and predicted improved mood and self-care on the subsequent day (Zajdel et al., 2018). When analyses of daily diary data were examined with the full sample (n = 207), we found that daily shared appraisal and daily collaboration were independently related to more supportive interactions and better mood for both patients and spouses, whereas daily collaboration was linked to better daily self-care (Zajdel et al., 2023). All of these reports are from an initial study of 207 couples in which one person was diagnosed with type 2 diabetes within the past 5 years, on average about 2 years ago (M = 1.88 years; SD = 1.68). Because that study focused on couples who were relatively new to their diagnosis of type 2 diabetes, the question remains as to whether communal coping changes over time and whether communal coping continues to have benefits as couples endure the disease for longer periods of time.

We followed these same couples for five years in order to address these questions. First, we asked whether communal coping changes over time. We had two competing hypotheses. One possibility is that couples’ initial response to a diabetes diagnosis is to come together cognitively (shared appraisal) and behaviorally (collaboration) to handle the disease, but that communal coping dissipates over time as couples habituate to the disease and its management. Alternatively, couples may learn to communally cope over time, perhaps in recognition that the disease and its effects persist—and often progress—and also in response to individual coping resources becoming depleted. To our knowledge, researchers have not investigated how communal coping changes over time across the disease process.

Second, we asked whether the relation of communal coping to health outcomes for the PWD changes over time. In the initial study, we documented benefits of communal coping, as described above. Do those benefits persist, decrease, or increase with time? Communal coping may be more important early in the disease process as there is greater uncertainty about how to respond and how the disease will progress. However, communal coping also may become more important later in the disease process as the novelty of the disease wears off and the PWD feels burdened by having to continue to manage the disease.

To address these questions, we employed a measurement burst design, comparing communal coping in the context of two 14-day diary assessment periods: one during the initial study and then the second during the follow-up study five years later. The present report compares communal coping across the two daily diary assessments as well as the links of communal coping to mood, coping efficacy, relationship outcomes (support, perceived responsiveness), and self-care (blood glucose checking, blood glucose values, dietary adherence) for PWD across the two assessments.

Method

Transparency and Openness

All data, analysis code, and research materials are publicly available at https://osf.io/5dbxs/. Data were analyzed using SPSS.

Participants and Procedure

We obtained Carnegie Mellon and University of Pittsburgh Institutional Review Board approval. Couples were recruited from the community via health fairs, mass media advertising, and brochures from physician offices. Interested persons contacted the project director by phone and were screened for eligibility. Eligibility requirements were: type 2 diabetes diagnosis within the past 5 years, no other illness that affected daily life more than diabetes (e.g., cancer), and married/cohabiting with a partner who did not have diabetes and was willing to participate in the study. Additional recruitment details are provided in Helgeson et al., (2019). This study is novel in comparison to other research on type 2 diabetes which typically enrolls couples without regard to length of diagnosis (which is typically years if not decades); here we capture PWD at a relatively recent phase in the disease process.

We enrolled 207 couples in the first study (Time 1) between September 2012 and December 2017. Although couples were enrolled in the study, the present paper is based on data only from the PWD, as the hypotheses center on whether and when the PWD benefits from communal coping. The demographics of this initial sample are shown in Supplementary Table 1. Couples were met by two experimenters either in their homes (71.5%) or at the university (28.5%). First, informed consent was obtained from both individuals. Then, couple members completed surveys and were interviewed independently, followed by a videotaped discussion about diabetes (this discussion is not included in the present investigation). At the end of the session, each couple member was asked to complete a brief survey at the end of the day for the next 14 days. Daily surveys contained questions about communal coping, mood, social interactions, and, for PWD, self-care. To facilitate questionnaire completion, we distributed internet-connected tablets to patients and partners. Daily diary compliance was high, with the PWD completing an average of 12.33 (SD = 1.66) out of 14 surveys. Participants were paid $50 for the in-person session and $100 for the daily diary portion of the study.

Five years later we recontacted couples to conduct a similar follow-up session (Time 2). We retained 133 of the PWD. Reasons for attrition included death of PWD (n = 9), participation refusal (n = 10), unable to contact (n = 28), contacted but couldn’t schedule (n = 23), and other circumstances (n = 4; 2 couple members arrested, 1 patient had an abusive partner, 1 partner unexpectedly died). Of the 133 PWD who participated at Time 2, 108 of those had partners. Of the 25 without partners, 12 had broken up, 8 partners had died, 5 partners refused. Of the 108 with partners, 3 participants did not complete Time 1 daily diaries, 3 participants who completed Time 2 daily diaries did not complete Time 1 daily diaries, and 3 partners were new at Time 2 (which meant we could not compare over time). Demographics for the final sample of 99 PWD that we examined at both Time 1 and Time 2 are shown in Table 1.

Table 1.

Sample Demographics for Person with Diabetes (n = 99)

Variable
Gender*	61% male 39% female
Race	64% White 36% Black
Ethnicity	95% non-Hispanic 5% Hispanic
Education	2% <HS 27% HS grad 12% some college 28% 2 year grad 17% 4 year grad 13% post-grad M=3.71 (SD = 1.42)
Work status	41% no 59% yes
Work hours	M=40.32, SD=13.35
Income	Median range $50-$59,000
Age	M=53.54 SD=11.26
Marital status	82% married 18% unmarried
Gender composition of the couple	97 different gender 2 same gender
HbA1c	M=6.78, SD=1.22
Diabetes duration	M=1.81, SD=1.68

^*At the time the data were collected, we asked participants to identify their gender and did not distinguish between sex and gender identity.

The follow-up session paralleled the initial session and was conducted between December 2017 and July 2021. Couple members completed surveys, were interviewed separately, and participated in a videotaped discussion about diabetes. The follow-up session was conducted in-person until the COVID-19 pandemic emerged. After stay-at-home orders were issued on March 16^th, 2020, we conducted the remaining interviews (23%) remotely. (However, note that the data for the present report are the daily diary data, which were collected remotely at both Time 1 and Time 2.) At the end of the session, couple members again completed 14 daily surveys. We distributed internet-assisted tablets to couple members who did not have their own devices. Again, participants were paid $50 for the in-person session and $100 for the daily diary portion of the study. Daily diary compliance was high, with the PWD completing an average of 13.42 (SD = 1.13) out of 14 surveys.

Instruments

The means and standard deviations for the aggregate of all the daily diary variables at Time 1 and Time 2 are shown in Table 2 along with their reliabilities. We used variance components analysis to establish the within-person reliability for scales with at least 3 items (Bolger & Laurenceau, 2013). When scales had only two items, we used traditional multi-level modeling to examine the relation between the two items.

Table 2:

EMA Variables for Follow-up Participants at Time 1 and Time 2 (n = 99); Aggregate Mean (Standard Deviation) and Variance Component Analysis

	Time 1		Time 2
	M (SD)	Reliability (VCA or Beta)	M (SD)	Reliability (VCA or Beta)
Appraisal	1.84 (.79)	n/a	1.51 (.73)	n/a
Collaboration	2.40 (1.33)	n/a	2.05 (1.25)	n/a
Happiness	3.77 (.73)	.72	3.57 (.87)	.81
Depression	1.44 (.51)	.77	1.33 (.46)	.75
Anger	1.54 (.51)	.76	1.34 (.40)	.79
Anxiety*	1.55 (.58)	.54, p <.001	1.42 (.62)	.59, p < .001
Coping Efficacy	3.84 (1.07)	n/a	3.27 (1.20)	n/a
Emotional support*	2.59 (.81)	.43, p < .001	2.46 (1.00)	.55, p < .001
Instrumental support*	1.91 (.73)	.57, p < .001	1.98 (.92)	.65, p < .001
Unsupportive interactions	1.17 (.30)	.67	1.15 (.36)	.67
Perceived positive response*	.01 (.83)	.77, p < .001	.02 (.85)	.84, p < .001
Perceived negative response*	−.04 (.68)	.23, p < .001	−.02 (78)	.27, p < .001
Blood glucose checking	.65 (.38)	n/a	.52 (.41)	n/a
Dietary adherence	3.48 (.81)	n/a	3.57 (.71)	n/a

Note: n/a: only one item was used to assess the construct;

^*these are two-item scales, so the reliability is reported as a beta coefficient of one variable predicting the other with the p value noted.

Communal coping.

The appraisal aspect of communal coping was measured by one item: When you thought about diabetes today, did you view it as ‘our problem’ (shared by your and your partner equally) or mainly your own problem? (1) completely my problem, (2) mostly my problem, (3) both of our problem. The collaboration aspect of communal coping was measured with the item: How much did you and your partner work together to take care of diabetes today? (1=none of the time; 5 = all of the time). We used abbreviated measures of communal coping (and all measures) to facilitate compliance with daily completion. These single item measures have been used elsewhere (e.g., Berg et al., 2008; Helgeson et al., 2022; Stephens et al., 2013) and are the central items from longer scales.

Mood and Coping Efficacy.

We used the 3-item happy, depressed, and anxiety subscales from the Profile of Mood States (Usala & Hertzog, 1989). Three face valid items were used to measure angry mood: angry, annoyed, mad (Zajdel et al., 2018). Because the reliability of the anxiety scale was problematic at Time 1, we replaced the item at Time 2. However, for comparison purposes, we could only use the two items that were the same across studies: anxious and nervous. Because we were concerned about the number of dependent measures, we examined the intercorrelation of the four mood measures at the aggregate level. Depression, anxiety, and anger were strongly correlated (r’s ranged from .69 to .81 in Time 1 and .66 to .75 in Time 2). Thus, we standardized the three scale scores and took the average to create a negative affect index. Happy was retained as the positive mood variable. To measure coping efficacy, we also had a one-item coping question in which we asked participants how they were dealing with diabetes on a 5-point scale, ranging from very poor to very well (Berg et al., 2008).

Relationship outcomes.

We measured emotional support, instrumental support, and unsupportive interactions with the scale developed for this study and described in detail (including reliability and validity data) in Helgeson et al., (2017). As described in Helgeson et al. (2017), the emotional support items were taken from Fekete et al.’s (2007) emotional support scale; instrumental support and unsupportive interactions were measured with items from Schaefer et al.’s Diabetes Family Behavior Checklist (Schafer et al., 1983). We also measured both perceived positive and negative emotional responsiveness with items from Fekete et al., (2007). PWD were asked to think about how their spouse responded to them with respect to diabetes—how much they felt understood, supported, judged, and ignored. In both cases, the first two items represented positive perceived responsiveness, and the second two items represented negative perceived responsiveness. Response scales for each item ranged from 1 (not at all) to 4 (a lot). We examined the intercorrelation of all five scales at the aggregate level to see if we could create composite indices. Emotional support, instrumental support, and perceived positive emotional responsiveness’s intercorrelations were .54, .73, .75 at Time 1 and .63, .79, .81 at Time 2. Thus, we created a positive support index by taking the average of the three standardized scales. Unsupportive interactions and negative perceived emotional responsiveness were correlated .56 at Time 1 and .55 at Time 2, so we averaged these two standardized scales to create a negative support index.

Self-care outcomes.

We had three measures of self-care. First, we asked PWD if they had checked their blood glucose that day (0=did not test, 1=did test). If they indicated yes to glucose checking, we asked them to report their most recent glucose value for that day. Note that not all patients check their blood glucose on a daily basis. In this sample, 80% of the participants responded to the question at least once. Third, we measured dietary adherence with the question, “How much did you follow your diet today?” Responses were made on a 5-point scale ranging from 1 = not at all to 5 = a lot. These brief measures of self-care were selected to facilitate diary compliance. They are face valid items, some of which have been used in other diabetes research (Helgeson et al., 2024).

Overview of the Analysis

We addressed differential attrition by conducting a series of independent t-tests to compare those retained in the study to those lost to follow-up. Then, we examined whether communal coping changed between Times 1 and 2 with paired t-tests, examining appraisal and collaboration separately.

Prior to examining our primary research questions, we examined whether there were any demographic or background variables that needed to be statistically controlled because they could confound the relation between communal coping and outcomes. Then, to examine whether collaboration and appraisal predicted outcomes at both Time 1 and Time 2, we used multi-level modeling with restricted maximum likelihood and entered collaboration (or appraisal), time, and the interactions with time in the model to predict outcomes. Research shows that the components of communal coping (collaboration and shared appraisal) are distinct factors with independent links to outcomes (Zajdel & Helgeson, 2020). Thus, we evaluated collaboration in one set of analyses and appraisal in another set of analyses. For each, we parsed the between-participant (BP) and within-participant (WP) variance by using the person-centered mean to represent the BP variable and subtracting each person’s average from the daily score for the WP variable. A BP effect shows that people who score high on the independent variable differ from those who score low on the independent variable (e.g., people with higher shared appraisal are happier than people with lower shared appraisal). A WP effect shows that on days in which a person scores high on an independent variable differ from days in which that same person scores low on an independent variable (e.g., on days when people have a higher shared appraisal they are happier than on days in which they have a lower shared appraisal). Because glucose checking is a dichotomous outcome, we used MLM for logistic regression for this outcome. For each predictor/dependent variable pair we present two models, as shown in Tables 3, 4, and 5. In the first model, we show the main effects of the BP and WP collaboration/appraisal and time. In the second model, we add the interactions of collaboration/appraisal with time.

Table 3:

Multi-Level Models: Relations of Collaboration and Appraisal to Mood and Coping

	Happy		Negative mood		Effective Coping
Collaboration
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
age	.01(.01); p=.276	.01(.01); p=.269	−.01(.00); p=.110	−.01(.00); p=.108	−.00(.01), p=.990	.00(.01); p=.930
education	−.01(.05); p=.812	−.01(.05); p=.809	.01(.03); p=.861	.01(.03); p=.855	−.07(.04); p=.085	−.08(.04); p=..072
income	.03(.03); p=.265	.03(.03); p=.262	−.05(.01); p<.001	−.05(.02); p<.001	.01(.02); p=.593	.01(.02); p=.498
BP	.32(.08); p<.001	.28(.08); p=.001	−.14(.05); p=.004	−.16(.05); p=.001	.36(.07); p<.001	.26(.07); p<.001
WP	.06(.02); p=.006	.06(.03); p=.026	−.01(.02); p=.484	−.01(.02); p=.810	.19(.03); p<.001	.12(.04); p<.001
time	−.18(.04); p<.001	−.18(.04); p<.001	−.15(.03); p<.001	−.15(.03); p<.001	−.53(.05), p<.001	−.53(.05); p<.001
BP X time		.09(.05); p=.050		.05(.03); p=.074		.22(.05); p<.001
WP X time		.00(.04); p=.977		−.01(.02); p=.572		.13(.04); p=.002
Appraisal
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
income	.02(.03); p=.432	.02(.03); p=.429	−.05(.01); p=.001	−.05(.01); p=.001	−.01(.02); p=.779	−.01(.02); p=.783
BP	.37(.13); p=.005	.28(.13); p=.037	−.21(.07), p=.005	−.23(.08); p=.003	.30(.12); p=.011	.19(.12); p=.131
WP	.10(.05); p=.024	.06(.06); p=.296	−.03(.03); p=.250	−.05(.04); p=.156	.06(.06); p=.317	.17(.07); p=.015
time	−.19(.04); p<.001	−.19(.04); p<.001	−.14(.03); p<.001	−.14(.03); p<.001	−.56(.05); p<.001	−.57(.05); p<.001
BP X time		.17(.07); p=.017		.04(.04); p=.342		.27(.08); p=.002
WP X time		.11(.07); p=.139		.04(.05); p=.416		−.28(.09); p=.003

Note: Time is scored as 0 = Time 1; 1 = Time 2; BP = between person; WP = within-person

Table 4:

Multi-Level Models: Relations of Collaboration and Appraisal to Relationship Outcomes

	Positive Support Index		Negative Support Index
Collaboration
	Model 1	Model 2	Model 1	Model 2
age	.01(.00); p=.214	.01(.00); p=.219	.00(.01); p=.727	.00(.01); p=.714
education	−.00(.03); p=.937	−.00(.03); p=.978	−.02(.04); p=.672	−.02(.04); p=.649
income	.00(.02); p=.999	−.00(.02); p=.931	−.05(.02); p=.020	−.05(.02); p=023
BP	.58(.05); p<.001	.54(.05); p<.001	−.02(.07); p=.828	−.04(.07); p=.631
WP	.23(.02); p<.001	.27(.02); p<.001	−.00(.02); p=.878	−.03(.03); p=.232
time	.07(.03); p=.018	.07(.03); p=.013	−.01(.03); p=.881	−.00(.03); p=.892
BP x time		.08(.03); p=.021		.04(.04); p=.267
WP x time		−.09(.02); p<.001		.06(.03); p=040
Appraisal
	Model 1	Model 2	Model 1	Model 2
income	−.02(.02); p=.265	−.03(.02); p=.261	−.07(.02); p=.002	−.07(.02); p=.002
BP	.55(.11); p<.001	.42(.11); p<.001	−.22(.10); p=.029	−.24(.11); p=.026
WP	.29(.04); p<.001	.31(.04); p<.001	−.05(.03); p=.155	−.14(.04); p=.002
time	.09(.04); p=.008	.09(.03); p=.010	−.01(.03); p=.677	−.01(.03); p=.758
BP x time		.25(.06); p<.001		−.01(.06); p=.930
WP x time		−.04(.05); p=.348		.21(.06); p<.001

Time is scored as 0 = Time 1; 1 = Time 2; BP = between person; WP = within-person

Table 5:

Multi-Level Models: Relations of Collaboration and Appraisal to Diabetes Self-Care

	BG Checking		BG Reading		Dietary Adherence
Collaboration
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
age	−.01(.02); p=.515	−.01(.02); p=.529	−.01(00); p=.048	−.01(.00); p=.035	−.00(.01); p=.799	−.00(.01); p=.763
education	−.42(.17); p=.019	−.42(.17); p=.019	.01(.02); p=.564	.01(.02); p=.498	−.02(.05); p=.685	−.02(.05); p=.677
income	.09(.09); p=.342	.09(.09); p=.337	−.00(.01); p=.843	−.00(.01); p=.772	.05(.02); p=.039	.05(.02); p=.042
BP	.41(.31); p=.188	.21(.34); p=.537	−.01(.03); p=864	.05(.03); p=.122	.24(.08) p=.003	.30(.08); p<.001
WP	.13(.08); p=.107	.14(.09); p=.144	−.02(.01); p=.188	−.02(.02); p=.271	.10(.02); p<.001	.10(.03); p=.002
time	−.85(.31); p=.018	−.83(.31); p=.019	.03(.02); p=.241	.03(.02); p=.140	.10(.04); p=.022	.10(.04); p=.024
BP x time		.36(.29); p=.220		−.11(.02); p<.001		−.12(.05); p=.010
WP x time		−.02(.12); p=.846		.00(.02); p=.862		.00(.04); p=.993
Appraisal
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
income	−.01(.09); p=.939	−.01(.09); p=.930	−.01(.01); p=.488	−.01(.01); p=.515	.04(.02); p=.093	.04(.02); p=.091
BP	.06(.47); p=.900	−.77(.56); p=.172	−.03(.05); p=.551	.04(.05); p=.418	.22(.12); p=.063	.30(.12); p=.018
WP	−.14(.14); p=.332	.03(.16); p=859	−.02(.04); p=.662	.02(.04); p=.690	.13(.05); p=.007	.17(.06); p=.003
time	−.94(.32); p=.014	−.96(.32); p=.009	.02(.02); p=.409	.02(.02); p=.485	.10(.05); p=.024	.10(.05); p=.025
BP x time		1.51(.53); p=.011		−.12(.04); p<.001		−.14(.08); p=.057
WP x time		−.43(.25); p=.090		−.09(.05); p=.097		−.11(.08); p=.198

Time is scored as 0 = Time 1; 1 = Time 2; BP = between person; WP = within-person; BG = blood glucose

Missing data were minimal. Participants completed an average of 12.4 out of 14 and 13.4 out of 14 surveys during the Time 1 and Time 2 diary periods, respectively. Within the daily diary surveys, missing data was less than 1% for most variables. The exception to low missingness was the glucose reading variable, which was missing in 42% of cases. In this case missingness is expected because participants only respond to this question if they answered yes to the glucose checking item (i.e., data were missing at random). Thus, the restricted maximum likelihood estimator was used in all models to account for missing data under the assumption that data were missing at random (Little & Rubin, 2019). However, we note that models evaluating glucose reading have reduced power due to high levels of missingness.

Results

Differential Attrition

As mentioned previously, the demographics for the sample at Time 1 and Time 2 are shown in Supplementary Table 1. We compared those who dropped out of the study at Time 2 to those who were retained on these demographic variables. There were only two significant differences. Those retained at Time 2 had a higher education (M = 3.71, SD = 1.46) than those lost to follow-up (M = 3.22, SD = 1.35) at baseline, t(204) = 2.39, p < .05. Those retained at Time 2 also had a lower baseline HbA1c (M = 6.92, SD = 1.60) than those lost to follow-up (M = 7.67, SD = 2.03), t(205) = 2.94, p < .01.

We examined whether those who dropped out from the study differed from those who were retained in the study on the independent and dependent variables. Only one difference appeared. Those retained in the study reported better dietary adherence at baseline (M = 3.47, SD = .82) than those who lost to follow-up (M = 3.20, SD .82), t(198) 2.33, p < .05.

Covariate Selection

We examined whether age, gender, race, education, income, and duration of diabetes were related to collaboration and appraisal. To the extent this was the case, they could confound the relation of collaboration and appraisal to outcomes. We found that collaboration was related to older age (r = .02, p < .05), lower education (r = −.13, p <. 05), and lower income (r = −.08, p < .05). Shared appraisal was related to lower income (r = −.04, p < .05). Thus, age, education, and income were statistically controlled in analyses of collaboration, and income was statistically controlled in analyses of appraisal.

Communal Coping Change Over Time

When we compared the aggregate collaboration and appraisal variables at Times 1 and 2, we found a decrease in both. Overall collaboration decreased from Time 1 (M = 2.37, SD = 1.00) to Time 2 (M = 2.05, SD = .97), t(99) = 3.80, p < .001, d = .82, and overall shared appraisal decreased from Time 1 (M = 1.83, SD = .64) to Time 2 (M = 1.50, SD = .63), t(99) = 5.791, p < .001, d = .57.

Multi-Level Modeling: Mood

The multi-level models for the prediction of mood are shown in Table 3 with collaboration in the top half and appraisal in the bottom half.

Collaboration.

As shown in Model 1, collaboration was related to more happiness at the BP and WP levels. As shown in Model 2, collaboration did not interact with time at either level. Collaboration was related to lower negative mood only at the BP level, but no interactions of collaboration with time emerged.

Collaboration was related to effective coping at the BP and WP level. Both interacted with time. BP collaboration was related to more effective coping, but the relation was stronger at Time 2 (.48, SE = .07; p < .001) than Time 1 (.26, SE = .07; p < .001; Figure 1). Similarly, WP collaboration was more strongly related to effective coping at Time 2 (.26, SE = .04, p < .001) than Time 1 (.12, SE = .04; p < .001; see Supp Figure 1). At both levels, low levels of collaboration appeared especially problematic at Time 2.

Figure 1:

Between person (BP) collaboration is more strongly linked to effective coping at Time 2 than Time 1

Appraisal.

There were BP and WP effects of appraisal on happiness, such that more shared appraisal at both levels was related to greater happiness (see Model 1). As shown in Model 2, the BP effect was qualified by an interaction with time. Consistent with the collaboration interactions, BP appraisal was more strongly linked to happiness at Time 2 (.45, SE = .13, p < .001) than Time 1 (.28, SE = .13, p < .05; see Figure 2).

Figure 2:

Between person (BP) appraisal is more strongly linked to happy mood at Time 2 than Time 1.

There was a BP effect of appraisal on negative mood, such that shared appraisal was related to a lower negative mood. There was no interaction with time.

There was a BP effect for appraisal on effective coping in Model 1, such that greater shared appraisal was related to more effective coping. Model 2 showed that there were BP and WP interactions with time. Consistent with the other interactions, BP appraisal was unrelated to effective coping at Time 1 (.19, SE =.12, p = .13) but related to more effective coping at Time 2 (.45, SE = .12, p < .001; see Supp Figure 2). The WP interaction, however, showed a different pattern. Greater shared appraisal was related to more effective coping at Time 1 (.17, SE = .07, p < .05) but was unrelated to effective coping at Time 2 (−.11, SE = .08, p = .15; see Figure 3).

Figure 3:

Within person (WP) appraisal is related to more effective coping at Time 1 but is unrelated at Time 2.

Multi-level Modeling Support

The multi-level models for the prediction of support are shown in Table 4, with collaboration at the top and appraisal at the bottom.

Collaboration.

BP and WP collab were related to greater positive support. Both effects were moderated by time. Again, BP collaboration was more strongly linked to positive support at Time 2 (.62, SE = .05, p < .001) than Time 1 (.54, SE = .05, p < .001; see Supp Figure 3). WP collaboration, by contrast, was stronger at Time 1 (.27, SE = .02, p < .001) than Time 2 (.18, SE = .02, p < .001; see Supp Figure 4).

There were no overall effects of collaboration on negative support, but there was a WP collaboration by time interaction in Model 2. Although it appears that WP collaboration was related to lower negative support at Time 1 (−.03, SE = .03, p = .23) but higher negative support at Time 2 (.03, SE = .03, p = .32), neither slope is significant see Supp Figure 5).

Appraisal.

Both BP and WP appraisal were related to greater positive support. The BP effect was qualified by an interaction with time. BP appraisal was more strongly related to positive support at Time 2 (.67, SE = .11, p < .001) than Time 1 (.42, SE = .11, p < .001; see Supp Figure 6).

Neither the BP nor the WP effects for appraisal were related to negative support. However, there was a WP effect moderated by time. Shared appraisal was related to lower negative support at Time 1 (−.14, SE = .04, p < .005) but was unrelated to negative support at Time 2 (.07, SE = .05, p = .14; see Supp Figure 7)—similar to the collaboration WP interaction.

Multi-Level Modeling Diabetes Self-Care (Table 5)

The multi-level models for the prediction of diabetes self-care are shown in Table 5, with collaboration at the top and appraisal at the bottom.

Collaboration.

There were no effects for BP or WP collaboration on blood glucose checking; nor were there interactions with time. There were no effects for BP or WP collaboration on glucose reading, but there was a BP by time interaction. BP collaboration was related to lower BG readings at Time 2 (−.07, SE = .03, p < .05) but was unrelated to BG readings at Time 1 (.05; SE = .03, p = .12; see Supp Figure 8). There were BP and WP effects of collaboration on diet in the predicted direction. The BP effect was qualified by an interaction with time, such that collaboration was more strongly related to greater dietary adherence at Time 1 (.30, SE = .08, p < .001) than Time 2 (.18, SE = .08, p < .05; see Supp Figure 9)—although both effects are in the predicted direction.

Appraisal.

There were no overall BP or WP appraisal effects on blood glucose checking, but the BP effect was moderated by time in Model 2, such that shared appraisal was related to less frequent checking at Time 1 (−.77; odds ratio .46, p = .17) but more frequent checking at Time 2 (.74; odds ratio 2.09, p = .18; see Supp Figure 10)—however, neither slope was statistically significant.

There were no BP or WP appraisal effects for BG reading, but Model 2 showed a BP appraisal interaction with time. BP appraisal was related to lower BG readings at Time 2 (−.08; SE = .05, p = .09) but was unrelated to BG readings at Time 1 (.04; SE = .05, p = .42; see Supp Figure 11), but neither slope was statistically significant.

The BP and WP effects of appraisal were related to greater dietary adherence but did not interact with time.

Discussion

This study assessed the implications of communal coping for health and well-being among people adjusting to type 2 diabetes using a measurement burst design. This design allowed us to probe different periods of time to understand how both daily (within-person) and between-person links shifted over a five-year period. Previous research has not examined how communal coping changes over time following stressor onset. This study showed that both the appraisal and collaboration components of communal coping decreased over time. Results also showed that the implications of communal coping shifted over time, and these shifts differed across levels of analysis. Specifically, person-level communal coping (i.e., BP effects) tended to be more strongly linked to positive outcomes at Time 2 than Time 1, but daily increases in communal coping over what one typically experienced (i.e., WP effects) tended to be more strongly linked to outcomes at Time 1 than Time 2. However, these patterns varied somewhat depending on the outcome in question.

Psychosocial Outcomes

Across Time 1 and Time 2, both person- and daily-level communal coping were linked to positive psychosocial outcomes. People who reported greater communal coping reported greater happiness, less negative mood, more effective coping, and more positive support from their partner compared to people who reported lower communal coping. Daily fluctuations in communal coping were less reliably linked to positive outcomes, but links emerged with happy mood, effective coping, and positive support. While we have previously reported main effects of communal coping in similar models among this sample (Zajdel et al., 2018; Zajdel et al., 2023), this research expands on previous work by incorporating the five-year follow up data, confirming that communal coping remains an important predictor of psychosocial well-being over time.

Importantly, we examined how the links of communal coping to psychosocial outcomes evolved over the five-year follow-up period. At the person-level, we found that communal coping was more strongly linked to happiness, effective coping, and positive support at Time 2 than Time 1. These links were generally consistent across both the collaboration and shared appraisal components of communal coping. For happiness and effective coping, these links were such that low levels of communal coping were particularly detrimental at Time 2. For positive support, high communal coping was particularly beneficial at Time 2. Thus, the links of person-level communal coping to psychosocial outcomes were strongest at Time 2.

Given that Time 1 took place relatively early in the disease course (M = 1.81 years after diagnosis), PWD at this stage may have been less knowledgeable of or less accustomed to the daily demands of diabetes management or may have been avoiding the reality of their disease. Partners, too, have much to learn in the face of a new diabetes diagnosis before being able to provide appropriate support. This learning curve suggests that people may learn to communally cope more effectively over time. Alternatively, as a progressive disease, diabetes may have been easier to manage earlier in the disease course. Independent management of diabetes may be more successful at Time 1, with communal coping offering a somewhat smaller advantage. At Time 2 when one’s disease is more challenging to manage, communal coping may become more important. This interpretation is consistent with the fact that HbA1c increased across the five years for individuals who participated in both waves of data collection (t = −3.00, p = .003; Time 1 M = 6.75, SD = 1.21; Time 2 M = 7.25, SD = 1.61). HbA1c is the primary clinical marker for disease state in diabetes, with lower numbers being more optimal. Thus, participants’ disease progressed, possibly requiring greater day-to-day care and attention and likely accompanied by higher disease-related distress (Fisher et al., 2010). It thus follows that coping communally with one’s partner was more important during the second wave when this disease burden was higher.

However, the within-person findings followed a different pattern. In general, these findings showed that daily communal coping was more strongly linked to psychosocial outcomes (effective coping, support, negative interactions) at Time 1 than Time 2—the inverse of what was found at the person-level. (However, we do note one exception to this pattern: within-person collaboration was more strongly linked to coping efficacy at Time 2 compared to Time 1.) It is important to note that this level of analysis is concerned with departures from an individual’s typical level of communal coping. It may be that early on in the disease course, coping patterns are in flux, so daily increases in communal coping are particularly helpful for psychosocial outcomes. However, after five years, a couple’s coping pattern is likely more established, and departures from this pattern may be less impactful. This interpretation is supported by the present data, as intraclass correlation coefficients (ICCs) increased from Time 1 to Time 2 for all outcomes, as well as for both components of communal coping. ICCs reflect the ratio of between-person variance to total variance; increases in ICCs indicate that people became more consistent in their coping and psychosocial well-being over the five-year period.

From this perspective, the contradictory findings at the within- versus between-person levels come to agreement, as both indicate the benefit of consistency in communal coping with diabetes. After five years, the best psychosocial outcomes occurred when couples consistently approached diabetes management communally (i.e., had higher person-level communal coping), while daily increases in communal coping yielded diminishing returns. We underscore the fact that communal coping was beneficial for most psychosocial outcomes across both time points; the interactions with time speak to the differences in the strength of these links.

Diabetes Outcomes

Overall, results showed more limited support for main effects of communal coping on diabetes outcomes. Collaboration (within- and between-person) and shared appraisal (within-person only) were only associated with dietary adherence. This may reflect the fact that the activities related to food and diet (e.g., meal prepping, shopping, cooking) are more amenable to sharing and thus more impacted by communal coping. Again, these findings largely align with past work on this sample showing mixed links to diabetes outcomes (Zajdel et al., 2023), but here we extend this work by incorporating data from the five-year follow-up.

Statistical interactions with time also showed more varied patterns. At the person level, both collaboration and appraisal were more strongly linked to lower (more optimal) glucose readings at Time 2 than Time 1, but in the case of appraisal neither simple slope was significant. Similarly, person-level appraisal was linked to higher probability of checking glucose at Time 2 (and possibly lower probability of checking glucose at Time 1), but again neither slope was significant. These findings are in the same direction as the psychosocial findings above, so may similarly be explained by couples either learning how to better communally cope with diabetes over time, or by the progressive nature of the disease rendering communal coping more important over time.

The pattern for diet, on the other hand, was in the opposite direction: person-level collaboration was more strongly linked to good diet at Time 1 compared to Time 2. While we don’t wish to overinterpret this inconsistent finding, it may speak to individuals’ habituation to diabetes over time and declining motivation to adhere to one’s recommended diet. Replication is needed to confirm and better understand this link. Finally, there were no daily-level interactions with time, indicating that daily fluctuations in communal coping were not differentially linked to diabetes outcomes over time. Thus, while the diabetes outcome findings were at times aligned with psychosocial findings, there was less consistency, making interpretation more challenging.

Finally, it is worth underscoring that overall levels of communal coping decreased across the two waves—even as person-level communal coping became more beneficial for health outcomes. This highlights a possible point of intervention. Increasing diabetes-related communal coping among couples managing diabetes, or simply helping such couples sustain communal coping over time, may support positive health outcomes. Communal coping-based interventions could focus on instilling a shared appraisal of diabetes, perhaps by encouraging couples to consider past stressors that they appraised as shared, and help couples identify ways in which they could work together to manage diabetes (see proof of concept study by (Zajdel & Helgeson, 2024). Intervention efforts in this space should focus on fostering stable communal coping patterns among couples and may be particularly important for those who have been managing diabetes for longer.

The key strength of the present study is the use of a measurement burst design to examine the implications of communal coping across multiple time points. While communal coping has been examined at both the daily and person-levels in previous work, the current design allowed us to identify how coping changes dynamically across a five-year period. Findings also highlight the importance of attending to the level of analysis in psychological and health research: distinct patterns of change occurred at the daily- versus person-level. Additional study strengths include a community sample with balanced representation across race (White/Black), gender (male/female) and diverse educational and socioeconomic backgrounds, supporting the generalizability of these findings. However, these findings do not generalize to other races and ethnicities, and we had very few couples with same-gender partners. It will be important for future research to determine whether gender and/or race moderate the findings reported in this paper.

We also note several study limitations. Importantly, a significant portion of our sample was lost to attrition over the five years. This limits the generalizability of study findings and reduced our statistical power. As noted previously, for one of our outcome variables (blood glucose reading), we had significant missing data because not everyone checks their blood glucose regularly; thus, we had limited power to detect effects for this outcome. To enhance completion of daily diary entries, several of our measures consisted of single items which make it difficult to establish reliability and to capture the full construct. In addition, differential attrition limits the generalizability of these findings. Those who were retained in the study had higher education, better glycemic stability (i.e., lower HbA1c), and better dietary adherence. The fact that the participants retained in the study were advantaged in these ways likely limited the variability in outcomes which reduced our ability to detect significant effects. However, differential attrition also means we do not know if the benefits of communal coping—especially with respect to the increased advantages over time—would apply to those who may be at the greatest risk for poor outcomes.

As an observational study we cannot make causal conclusions based on these findings. While theory and prior research (Zajdel & Helgeson, 2024; Zajdel & Helgeson, 2021) support communal coping as a causal factor in health and well-being outcomes, we cannot rule out alternative causal explanations. Finally, we acknowledge that our own gender (female) and race (White) limits the findings in that our identities could have influenced the questions asked and the way that this study was designed to address these questions.

In sum, the present research employed a measurement burst design to understand how communal coping with type 2 diabetes was related to a range of psychosocial and diabetes outcomes over a five-year period. Communal coping was consistently linked to positive psychosocial outcomes, replicating past work and supporting communal coping as an important resource in the context of type 2 diabetes. Links of communal coping to diabetes outcomes were less consistent. Importantly, person-level communal coping became more strongly linked to positive outcomes over time suggesting that enhancing communal coping may be a target for intervention. By contrast, daily fluctuations in communal coping yielded smaller impacts on outcomes over time, but this pattern was less consistent when it came to diabetes outcomes compared to psychosocial outcomes.

Public significance statement:

Among persons with type 2 diabetes, there is evidence that communal coping (i.e., appraising diabetes as a shared problem and working with one’s partner to manage diabetes) is beneficial for health. In following couples from shortly after diagnosis to 5 years later, the benefits of communal coping increased over time suggesting that future intervention efforts might target communal coping.