Romantic Relationships Among Emerging Adults With and Without Type 1 Diabetes Free

Participants included 183 young adults, of whom 88 had type 1 diabetes (53 female and 35 male) and 95 did not have type 1 diabetes (51 female and 44 male). The average participant age was 27 years. More detailed demographic information is provided in Table 1.

Participants were recruited from a longitudinal study that began when they were a mean age of 12 years (17). At study start, there were 132 youth with and 131 youth without type 1 diabetes. At this 15-year follow-up, we retained 70% of the sample. A comparison of the two groups on baseline demographic variables revealed that we retained more females (76%) than males (63%) (χ²[1] = 5.42, P <0.05), and those we retained came from families with a higher social status (t[261] = 2.94, P <0.01), but there were no differences by race, age, or, among those with diabetes, baseline A1C.

The data reported here are from the final wave of data collection, which consisted of an online questionnaire and A1C data collection at a mean age of 27 years. The questionnaire assessed a variety of indices of psychological well-being for both groups, as well as diabetes-related health for those with type 1 diabetes.

We asked participants their current relationship status with the following response options: married, engaged, dating, single, divorced/separated, and widowed. We also asked respondents the sex of their partner.

For those in a romantic relationship (i.e., married, engaged, or dating), we administered the five-item Quality of Marriage Index by replacing “spouse” with “romantic partner” and “marriage” with “relationship” (e.g., “We have a good relationship”) (18). We chose this instrument because of its brevity, face validity, and applicability to relationships outside of marriage. The internal consistency was high (α = 0.96).

For participants with diabetes, we administered a survey to measure partner diabetes-specific supportive and unsupportive interactions. We measured emotional support with four items (e.g., “Was there for me by giving me his/her undivided attention”) (α = 0.83), instrumental support with four items (e.g., “Suggested things that might help me manage diabetes”) (α = 0.92), and unsupportive interactions with three items (e.g., “Argued with me about how to take care of diabetes") (α = 0.92). We used these items because they have been used extensively in previous research with people with diabetes (19,20).

We measured depressive symptoms with the 20-item Center for Epidemiologic Studies Depression Scale (21). Participants rated each item on a scale of 0 (rarely or none of the time) to 3 (all the time, 5–7 days) (α = 0.93). We also administered the four-item Perceived Stress Scale (22), which asked participants how often they felt or behaved a certain way on a scale ranging from 0 (never) to 4 (very often) (α = 0.80). Because the two scales were highly correlated (r = 0.73), we standardized them and took the average to form a psychological distress index. We note that we found identical results for each of the two variables, as reported below.

We used the Satisfaction With Life Scale (23), which asked participants the extent to which they agreed or disagreed with five statements on a scale of 1–7 (α = 0.91).

We used the Purpose in Life Scale (24), which asked participants to indicate their agreement with each of seven items on a scale of 1–6 (e.g., “I have a sense of direction and purpose in life") (α = 0.77).

The eight-item version of the UCLA Loneliness Scale (25) was administered, in which respondents indicated how often they have experienced each feeling on a scale from 1 (never) to 4 (often) (α = 0.82).

We measured participants’ confidence in their abilities to engage in specific diabetes self-management behaviors with the self-efficacy subscale of the Multidimensional Diabetes Questionnaire (26). This seven-item measure asks participants to indicate on a scale of 0–100 how confident they are in managing various aspects of diabetes (e.g., “follow diet” and “check blood glucose regularly”) (α = 0.84). We then developed seven parallel items to measure outcome expectancies, or the likelihood that the participant would engage in each of these behaviors in the next 2 weeks, using the same 100-point scale (with 0 = 0% or no chance that you will engage in this behavior and 100 = 100% certain that you will engage in this behavior) (α = 0.87). Because the two were highly related (r = 0.86), we averaged the two to form an efficacy index.

We administered the Self-Care Inventory (27,28), which was updated by adding eight more contemporary items, as described previously (29). Respondents indicated how well they followed their physician’s recommendations on a five-point scale (from 1 = never to 5 = always) over the past 2 weeks for glucose checking, insulin administration, diet, exercise, and other diabetes behaviors regarded as important by the American Diabetes Association (α = 0.83).

We mailed the CoreMedica HemaSpot-SE self-collection test kit to participants and provided written instructions on how to use it. Participants mailed the kits to the laboratory for processing, and we received the results. Of the 88 participants, 81 returned these kits (92%).

First, we examined whether there were group and sex differences in the presence of a romantic relationship with χ² analysis and in relationship quality with ANOVA (Table 2). Second, we used hierarchical regression analysis to examine the link between relationship status and outcomes. We entered relationship status on the first step, the two-way interactions of relationship status with group and with sex on the second step, and the relationship status by group by sex interaction on the third step. Third, we conducted a similar regression analysis to examine the link between relationship quality and outcomes. Finally, for those with diabetes, we entered supportive diabetes interactions (emotional and instrumental) and unsupportive diabetes interactions into a simultaneous regression to predict psychological and diabetes outcomes, including interactions with sex. Tables 3–5 present the results of these regressions analyses, including standardized betas, which can be considered effect sizes.

Demographic data for both groups are shown in Table 1. There were no group differences in education, age, work status, or living situation. Those without diabetes reported a higher income than those with diabetes. The findings below are reported without income as a covariate because results remained unchanged when income was included in the model. The correlation matrix of all study variables is shown in Table 2.

All relationships were heterosexual for participants with diabetes, and 94% were heterosexual for those without diabetes. There was no overall group difference in the presence of a romantic relationship. The majority of both groups (76% without diabetes and 66% with type 1 diabetes) were in a romantic relationship (χ²[1] = 2.17, P = 0.14). However, the group effect was qualified by an interaction with sex. There were no sex differences in the presence of a romantic partner for those without diabetes; 77% of females and 75% of males were in a relationship (P = 0.87). However, females with type 1 diabetes were much more likely to be in a romantic relationship than males with type 1 diabetes (79 vs. 46%, χ²[1] = 10.55, P <0.001). Males with type 1 diabetes were less likely to be in a romantic relationship than the other three groups.

There were no group or sex differences in relationship quality.

As shown in Table 3, those in a romantic relationship had less loneliness and more life satisfaction, but relationship status did not predict diabetes outcomes. Neither group nor sex moderated these links. Those with diabetes reported more distress and loneliness, lower life satisfaction, and less purpose in life than the control group. Some, but not all, of these differences were included in an earlier report on this sample when individuals were a mean age of 18 years (30).

For those in a romantic relationship, Table 4 shows the results for the prediction of outcomes based on the quality of the relationship. There were group × relationship quality interactions on psychological distress, loneliness, and life satisfaction. Higher relationship quality was related to lower distress for those with diabetes (β = −0.55, P <0.001) but not for control subjects (β = −0.14, P = 0.25). Higher relationship quality was also related to reduced loneliness for those with diabetes (β = −0.59, P <0.001) but not for control subjects (β = −0.16, P = 0.18). Additionally, higher relationship quality was related to higher life satisfaction for those with diabetes (β = 0.49, P <0.001) but not control subjects (β = 0.15, P = 0.20). Relationship quality was related to greater purpose in life for the entire sample. There were no sex × group × relationship quality interactions.

Also as shown in Table 4, for diabetes outcomes, there was a main effect of relationship quality on self-efficacy such that higher relationship quality was related to higher self-efficacy. There were trends for relationship quality to be related to better self-care and lower A1C, but neither effect was statistically significant.

When emotional, instrumental, and unsupportive diabetes interactions were entered into the regression equation to predict outcomes for the type 1 diabetes group, unsupportive interactions predicted greater psychological distress, more loneliness, and a lower sense of purpose in life (Table 5). Unsupportive diabetes interactions predicted all three diabetes outcomes: lower efficacy, poorer self-care, and worse A1C. Diabetes instrumental support interacted with sex such that partner instrumental support was related to greater life satisfaction for males (β = 0.78, P <0.01) but not females (β = 0.05, P = 0.82). There was a marginal interaction between unsupportive interactions and sex such that unsupportive interactions were related to lower life satisfaction for males (β = −0.65, P <0.005) but not females (β = −0.20, P = 0.24).

Our first study goal was to examine whether emerging adults with and without type 1 diabetes were equally likely to be in a romantic relationship. Results indicated that males with type 1 diabetes were significantly less likely to be in a romantic relationship than females with type 1 diabetes or than male or female control subjects, despite the fact that all groups were the same age. This research is consistent with an earlier report on this sample that showed support from romantic partners was stable across the ages of 17–23 years for all groups except males with type 1 diabetes, for whom it declined. Taken collectively, both findings suggest that males with type 1 diabetes may have more difficulties establishing romantic relationships. However, once in a relationship, it is not the case that relationship quality suffers; neither group nor sex was linked to relationship quality.

The second study goal was to link relationship status and relationship quality to psychological and diabetes-related outcomes. For psychological outcomes, relationship status was linked to less loneliness and greater life satisfaction. These findings are consistent with a large body of research that shows romantic relationships, in particular marriage, are good for health (31). However, the vast majority of previous research is conducted on middle-aged or older adults who are married; this study shows that the presence of a romantic relationship is linked to good psychological health among young adults in the formative stages of relationship development.

Additional analysis showed that the quality of the relationship was connected to better psychological well-being for those with type 1 diabetes but not for those without diabetes. This finding is consistent with previous research that showed relationships (specifically, conflictual interactions) were more strongly related to health for adolescents with diabetes than for those without diabetes (32). Taken collectively, these two studies suggest that both the positive and negative aspects of relationships are more impactful for those with than for those without diabetes.

In terms of diabetes outcomes, relationship status alone did not predict health, but the quality of the relationship mattered. Higher-quality relationships were connected to higher self-efficacy, and there were trends toward better self-care and A1C. When we examined the quality of the relationship in more detail among those with type 1 diabetes by focusing on supportive and unsupportive diabetes interactions, we found that unsupportive interactions were connected to two of the three psychological outcomes and all three of the diabetes-related outcomes, whereas supportive interactions were not. These findings are consistent with other research that has shown that negative interactions with network members are sometimes more harmful than positive interactions are helpful (33).

Taken collectively, these findings have implications for how practitioners can help young adults navigate their diabetes in the context of early romantic relationships. The quality of romantic relationships matters for those with diabetes, and unsupportive interactions may be especially likely to detract from diabetes care. The unsupportive interactions examined here involved reports that partners criticized how they took care of diabetes, nagged them about taking care of diabetes, and argued with them about how they took care of diabetes. In this case, romantic partners may be overstepping in their efforts to help their partner with type 1 diabetes. Although these kinds of interactions are never pleasant, they may be especially detrimental in the early stages of a relationship, when how diabetes is negotiated between the two people has not yet been clarified. This is an area in which diabetes clinicians and educators can help. A systematic review of interventions aimed at emerging adults with type 1 diabetes concluded that benefits were inconclusive. However, it is noteworthy that none of these interventions focused on or involved romantic partners (34).

Because several of our findings were qualified by participants’ sex, this study shows the importance of considering people’s background characteristics when studying the link between relationships and health. Racial, ethnic, and cultural background variables are also important factors to take into consideration. The current study was limited in that the majority of participants were Non-Hispanic White and therefore could not examine racial or ethnic differences. However, relationships might have an even stronger impact on health among participants who come from more interdependent cultures or communities.

Before concluding, we acknowledge several study limitations. As noted previously, the generalizability of this sample is limited by the lack of ethnic and racial diversity. In addition, at the time this study was initially conducted in 2002–2004, we did not collect any information on gender identity—only biological sex. The gender binary assumption is a study limitation. The study is also cross-sectional, precluding our ability to draw causal inferences from relationship status or relationship characteristics to health. We do not know if relationship characteristics led to poor diabetes outcomes or if poor diabetes outcomes led to relationship difficulties. Future longitudinal work should examine the implications of relationship characteristics for subsequent psychological and physical health, as well as the implications of psychological well-being and physical health for the development of romantic relationships over time.

We also did not collect data on the length of romantic relationships, which is certainly likely to have an impact on their effect on health. Future research should examine whether relationship quality has differential effects on heath depending on the stage of the relationship. For example, unsupportive interactions, such as partner nagging or criticizing diabetes self-care, may be viewed differently in the context of a long-term committed relationship than the early stages of a dating relationship.

Finally, some participants were not retained in this final wave of data collection—most notably, emerging adult males with type 1 diabetes. The relationship status and relationship quality of those lost to follow-up is unknown.

This study had several strengths that enhance its contribution to the literature. First, we know very little about the romantic relationships of emerging adults with type 1 diabetes. Because individuals in this population are in transition from the pediatric to the adult health care system, they are a notoriously difficult group to track. Among the studies that do exist, romantic relationships have been a neglected topic of research. This study shows the importance romantic relationships may have not only on psychological health, but also on the physical health of young adults with type 1 diabetes. It is important for future research to track these relationships over time and to examine the implications that relationship stability and breakup have for diabetes management and well-being.

The author acknowledges the research assistants who were involved in data collection, Abigail Vaughn for her assistance with data analysis, and Fiona Horner for her comments on a previous version of this manuscript.

This work was supported by National Institutes of Health grant R01 DK060586.

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

As the sole author, V.S.H. oversaw data collection and drafted and edited the manuscript. She is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.