|Home | About | Journals | Submit | Contact Us | Français|
ObjectiveThe initial validation of a brief assessment of a diabetes-specific self-esteem dimension in adolescents with type 1 diabetes.MethodsYouths with type 1 diabetes (n = 87) aged 10–16 years were administered the multidimensional Self-Esteem Questionnaire (SEQ) and a newly designed assessment of diabetes-specific self-esteem (DSSE). Their parents completed parallel forms. Adherence to the diabetes regimen and glycemic control were also assessed.ResultsIn factor analysis, DSSE items formed a distinct dimension of self-esteem in addition to the SEQ dimensions. This factor uniquely contributed to differences in youths’ global self-esteem. Significant associations with adherence and glycemic control suggested its concurrent validity. Agreement between youth- and parent-report DSSE forms supported inter-rater reliability.ConclusionsThe findings provide preliminary support for recognizing the importance of a DSSE dimension in adolescents’ adjustment to diabetes, and for the reliability and validity of the proposed assessment strategy.
Self-esteem is widely assumed to play a vital role in children's and adolescents’ adjustment to chronic illness. It is broadly recognized both as an outcome variable—an index of mental health—and as a predictor and regulator of beneficial health behavior (Harter, 1999; Marsh & Craven, 2006). This dual role of self-esteem is particularly evident in youths’ adjustment to type 1 diabetes. The demands of living with diabetes pose chronic stress on young patients and their families. They are required to follow a complex and challenging treatment regimen consisting of daily administration of insulin, vigilant monitoring of diet and exercise, and meticulous correction of care based on the results of frequent blood glucose checks (Wysocki & Greco, 1997). Conventional wisdom suggests that this exceptional treatment burden may substantially impact a youth's self-esteem. At the same time, maintaining a positive sense of self may be an essential resource for sustained adherence to the treatment regimen and the consequent prevention of serious long-term complications (White et al., 2001).
Even though these suppositions have immediate face validity, they have received only scant empirical support. There is little evidence that living with diabetes affects adolescents’ self-esteem levels; with few exceptions (Rovet, Ehrlich, & Hoppe, 1987), most findings indicate that youths with diabetes on average score just as favorable on self-esteem and self-concept scales as their physically healthy counterparts (Grey, Cameron, Lipman, & Thurber, 1995; Hanson et al., 1990; Helgeson, Snyder, Escobar, Siminerio, & Becker, 2007; Kovacs et al., 1990). Furthermore, although some studies have reported significant links between high self-esteem and better adherence (Jacobson et al., 1987), as well as better glycemic control (Maharaj, Daneman, Olmsted, & Rodin, 2004), others have found these associations to be inconsistent and weak in magnitude (Bryden et al., 2001; Kovacs, Goldston, Obrosky, & Iyengar, 1992).
It might be concluded from these equivocal findings that the self-esteem construct is not substantially implicated in adolescent adjustment to diabetes. However, it is also the case that current understanding of the development and functioning of self-esteem in chronically ill youths is still constrained and fragmented. Despite frequent examination of self-esteem in pediatric populations, few research efforts have focused systematically and precisely on this topic. Most commonly, self-esteem was afforded a supplemental and largely interchangeable role in studies aiming at broader issues of psychosocial adjustment and psychopathology. For that reason, it has been assessed in many studies solely in global, and hence, undifferentiated terms. In addition, psychological adjustment to chronic illness assessed by means of self-esteem was often analyzed as being a stationary and uniformly negative response to adversity (Hanson & Onikul-Ross, 1990; Harper, 1991).
In contrast, contemporary social–cognitive theories on the development and functioning of the self emphasize the notion that the self-system is multifaceted and dynamically adaptive (Harter, 1999; Marsh & Craven, 2006). Each life transition (such as entering school and meeting new peers) has distinctive demands and expectations associated with it. Acting within a new life context, individuals create an image of what they aspire to be like (a “desired” self) and what they are actually like (an “actual” self) in relation to the particular standards it poses. Theory suggests that, to the extent to which the actual self meets or fails to meet the desired self, high or low self-esteem with respect to this life domain is established (DuBois, Felner, Brand, Phillips, & Lease, 1996). The cognitive advances necessary to engage in this comparison emerge in middle to late childhood, such that during early adolescent development youths acquire an increasingly rich and multifaceted profile of self-evaluations across the various areas of experience (Harter, 1999). These context-specific components of self-esteem serve a 2-fold function, in that they form the basis for global feelings of self-worth, and in that they guide efforts to engage in each context by providing expectancies about the outcomes of future performance.
The notion that individuals acquire new facets of self-knowledge upon exposure to new contexts of experience has become a cornerstone of the conceptualization and measurement of self-esteem and self-concept development in physically healthy youths. On the other hand, this perspective has not been systematically applied to self-esteem research in pediatric populations. Given the daily management demands posed by their illness and the related expectations by their parents and health care providers, youths with type 1 diabetes will acquire a sense of how they ideally want and should be like in taking care of their illness, and they will evaluate to what extent their actual qualities and achievements meet this desired self. Thus, in the process of adapting to their unique environment, youths may generate a unique illness-specific dimension of self-esteem as part of their multifaceted self-portrait.
This possibility implies a shift in perspective on how diabetes impacts on youth self-esteem: whereas the traditional focus has been on self-esteem impairment resulting from the illness (Hanson et al., 1990), acquiring favorable self-esteem in relation to the demands of diabetes may be distinctly important for the development of a positive sense of general self-worth in response to the illness. In addition, research on multidimensional self-esteem in healthy youths suggests that performance and achievement are most strongly guided by dimensions of self-esteem that directly pertain to the same settings and domains (Marsh & Craven, 2006). Thus, self-esteem specific to diabetes may be more strongly involved in the successful self-management of the illness and the maintenance of glycemic control than previously examined aspects of self-esteem.
Support for a multidimensional framework of self-esteem in physically healthy youths has been established in studies addressing the construct validity of measures designed for this purpose (Marsh & Craven, 2006). Consequently, the goal of the present study was the initial validation of a brief assessment of diabetes-specific self-esteem (DSSE) in youths with type 1 diabetes, designed as an extension of an extant multidimensional Self-Esteem Questionnaire (SEQ). Consistent with previous construct validation approaches and with the conceptualization described above, our first aim was to examine the role of DSSE within the multidimensional structures of self-esteem. We hypothesized that DSSE would represent a distinct dimension in addition to dimensions pertaining to healthy adolescent development, and that it would uniquely contribute to youth ratings of general self-worth. Our second aim was to examine the relationships between the self-esteem dimensions and diabetes self-care outcomes (adherence to the diabetes regimen and glycemic control). We hypothesized that DSSE would be more strongly associated with these outcomes than nondiabetes-specific dimensions of self-esteem. The third aim was to examine inter-rater reliability by comparing youths’ self-report DSSE ratings with those obtained from parents in a corresponding parallel assessment of their child's DSSE. We hypothesized that there would be at least moderate parent–child agreement of DSSE ratings.
Youth–parent dyads were recruited from a pediatric outpatient clinic at an urban university medical center. All patients diagnosed with type 1 diabetes for a minimum of 1 year and in the age ranges of 10–16 years were identified through a review of medical records. Patients were excluded from the study if they were diagnosed with an additional major chronic illness (except for well-controlled asthma or thyroid problems), or if youth or parent were non-English speaking/writing. Of 120 families contacted, 73% agreed to participate and completed the assessments. Incomplete data from one parent could not be used, resulting in a study sample of 87 youths and 86 parents. The youths had a mean age of 13.4 years (SD = 2.06), and 55% were girls (n = 48). In terms of racial/ethnic background, 74% (n = 64) were White, and 14% (n = 12) were African American. The average duration of diabetes was 6.3 years (range 1.0–15.5, SD = 3.83). Approximately 74% (n = 64) of the sample used multiple daily injections with a combination of short and longer acting insulin, and 26% (n = 23) used an insulin pump.
The SEQ (DuBois et al., 1996) was used to assess youth self-esteem in “generic” (i.e., nondiabetes-specific) life-domains. The SEQ was developed and validated with a large (N = 1,800) and demographically diverse sample of physically healthy youths. Construction of the questionnaire was based on a definition of self-esteem as the evaluation of one's actual self in comparison to one's desired self, or the degree of liking for and satisfaction with the self. Consistent with this definition, each item asks directly about the extent to which the youth is satisfied or dissatisfied with the described aspect of self (e.g., “I do as well on tests in school as I want to” or “I like my body just the way it is”). Items are rated on a four-point scale ranging from strongly disagree to strongly agree. The measure consists of 42 items, yielding scale scores of self-evaluation pertaining to each of five separate life domains, that is, peer relations (eight items), school achievement (eight items), family (eight items), body image (four items), and sports/athletics (six items). A separate eight-item global self-esteem scale measures overall perceptions of self-worth (e.g., “I am happy with myself as a person”). The SEQ has demonstrated strong internal consistencies and test–retest reliabilities of its subscales, as well as factorial validity of its underlying multidimensional, hierarchical structure (DuBois et al., 1996; DuBois, Tevendale, Burk-Braxton, Swenson, & Hardesty, 2000). Evidence for consistent and theoretically expected relationships with indicators of physically healthy youth functioning is also supported (DuBois et al., 1996, 2002).
The Child Self-Esteem Questionnaire (CSEQ; DuBois et al., 1996) was used to obtain ratings of youth self-esteem from their parents. The CSEQ is an abbreviated 24-item parent-report version of the SEQ, with four items used to assess each of the five domains of self-esteem targeted on the SEQ, and four separate items for global self-esteem. Items parallel those on the SEQ, with wording changed to make them appropriate for parental rating (e.g., the item “I am happy with myself as a person” is changed to “This child is happy with himself/herself as a person”). The CSEQ scale scores have previously demonstrated strong internal consistency, as well as convergent and discriminant validity with self-report ratings on the SEQ (DuBois et al., 1996, 2002). In the present study, coefficient α's ranged from.78 to.90 for the SEQ subscales and from.83 to.91 for CSEQ subscales (Table IV).
A new set of items was constructed for the present study to assess youth self-esteem in the domain of diabetes care. In accordance with the conceptual framework underlying the SEQ, DSSE was defined as the degree of self-satisfaction or self-approval in relation to the medical and social management of diabetes in everyday life. The content for the DSSE items was derived from literature review and from semi-structured pilot interviews with youths with Type 1 diabetes aged 8–18 years and their parents. Nine families, recruited independently of the study sample, provided written informed consent to be interviewed separately on two occasions at their homes, and were asked about their views on living with and managing diabetes. A set of 25 sample items was initially generated from these interviews. In keeping with the scale format of the SEQ, each item inquired about the degree of satisfaction with self in diabetes-specific contexts, adopting evaluative phrases used in the SEQ as appropriate. Items were rated on a four-point scale ranging from strongly disagree to strongly agree. A parallel pool of 25 items for construction of a parent-report version for the assessment of their child's DSSE was also generated. The youth and parent-report items were tested for face validity during a second home visit with the pilot families. After completing the measures, both youth and parent were interviewed individually about their choices, and about item contents and phrasing. Items with awkward or ambiguous wording were refined based on their recommendations, and 18 items judged most appropriate and face-valid for both the youth- and parent-report DSSE forms were selected to be included in this study. For administration in the present study, the 18 items were embedded within the SEQ/CSEQ, with diabetes-specific items and other self-esteem items being alternated with one another to ensure consistency with the theoretical notion that DSSE is integrated with other dimensions of self-esteem.
Glycemic control was indexed via HbA1c levels obtained from medical records. HbA1c is an indicator of a patient's average blood glucose levels present over the preceding 3 months. For each youth, the most recent HbA1c assay was used in analyses, recorded within 3 months of the date of questionnaire administration. The mean HbA1c in the sample was 8.0% with a SD of 1.58, indicating reasonably good glycemic control overall, yet considerable variation between youths (Mortensen et al., 1998).
Parent and child individually completed a modified version of the Diabetes Self-Management Profile (DSMP; Harris et al., 2000; Iannotti, Nansel et al., 2006), a structured adherence interview. The original DSMP uses an open-ended format. In the modified version (Iannotti, Nansel et al., 2006), interview administration and scoring are standardized to facilitate administration by nonmedical interviewers and to make the instrument more suitable for interviewing younger youths separately from parents. The instrument includes 29 items assessing adherence in the areas of insulin-administration, meal planning, exercise, blood glucose testing, and self-care adjustments. It has sound psychometric characteristics and good internal consistency, with coefficient α's of.73 for child and.75 for parent report in the present sample. Validity of the interviews was evident in that child and parent reports showed agreement, (r =.48, p <.001) and were similarly closely associated with HbA1c (r = −.29; p =.01 and r = −.32, p =.003, respectively). For the purposes of this study, child and parent reports were averaged to form an overall adherence score in order to increase reliability (Holmbeck, Li, Schurman, Friedman, & Coakley, 2002). This also slightly increased the association with HbA1c (r = −.36, p =.001), suggesting an improvement in validity compared with individual scores.
Institutional Review Board approval was obtained from the institutions involved. Parents provided written informed consent and authorization for disclosure and use of health information from the child's medical records. Youths provided written informed assent. All assessments took place in the home or other convenient location selected by the parent. Youths completed the self-administered SEQ and DSSE items, and parents completed the self-administered parallel versions, with the option of having the items read to them. Trained interviewers administered the one-on-one adherence interview individually to parent and child. Each participant (parent and child) received $25 for completing the assessments.
Following initial item analyses, the first analytic step was to examine whether DSSE items would form a distinct dimension of youth self-esteem in addition to the five dimensions assessed by the SEQ. Common factor analysis using maximum likelihood estimation was employed to examine the factor structure of the jointly pooled set of DSSE items and domain-specific SEQ items. An oblique factor rotation (promax) was utilized because different dimensions of self-esteem are theoretically expected to be positively associated with one another (Harter, 1999); empirically, the SEQ factors have previously shown moderate intercorrelations ranging from.25 to.50 (DuBois et al., 1996). In the second step, demographic differences on DSSE scores were examined with correlations (age, illness duration) and t-test (gender). Demographic differences on general self-worth, adherence, and glycemic control were also examined to identify demographic factors that should be included in the subsequent regression analyses to control for possible confounds. Step 3 examined whether DSSE would uniquely contribute to ratings of general self-worth, and step 4 investigated associations with adherence and with glycemic control; hierarchical multiple regression analysis was employed in both steps. Demographic variables were entered first, followed by the set of content-specific SEQ factors, and DSSE entered last to estimate the incremental variance proportion it accounted for. We also tested the goodness-of-fit of a constrained model in which the regression weight of DSSE was set to zero in comparison with the “saturated” (perfectly fitting) model that includes regression weights from all predictors; rejection of the constrained model indicates that the influence of DSSE needs to be taken into account for the model to fit the data. To reduce the number of correlated predictors in each regression analysis, only those predictor variables demonstrating a significant zero-order correlation with the criterion variable were included in any given model. Finally, step 5 of the analyses addressed agreement between youth self-report and parent-report DSSE forms using paired-sample t-tests and intraclass correlations. Because these are preliminary analyses of DSSE, the significance level was set at.05 for all analyses.
Our first aim was to derive a psychometrically sound yet brief instrument including no more than 10 items from the pool of 18 DSSE items. Since weakly correlated items have low communality and will, thus, likely perform poorly in factor analysis, we inspected the initial communalities of the items (the highest correlation and the squared multiple correlation of each item with the remaining items, see Floyd & Widaman, 1995). Eight items with the lowest communalities were eliminated, and 10 DSSE items (shown in Table I) were retained for subsequent analyses.
We then submitted the DSSE items together with the 34 domain-specific SEQ items to common factor analysis. Inspection of the scree plot suggested that six factors should be retained. Goodness-of-fit of the rotated six-factor solution was nonsignificant [χ2 (697) = 750.82; p =.08], indicating an adequate fit. As expected, the solution replicated the five dimensions of healthy adolescent self-esteem: items of each of the SEQ subscales had substantial loadings on the factor they were intended to assess (average item loadings ranging from.58 to.67) and no significant cross-loadings (see lower section of Table I). The additional, sixth, factor was comprised exclusively of diabetes-specific items. All DSSE items had loadings on this factor >.45 (with an average of.63) and no sizeable cross-loadings on any other factor (see upper section of Table I). As shown in Table II, correlations between the DSSE factor and the domain-specific SEQ factors were moderate, ranging from.27 (p =.01) to.56 (p <.001).
For comparison, alternative factor solutions were also inspected. A five-factor solution recovered four SEQ factors and a diabetes-specific factor. However, items from the SEQ body image subscale did not significantly load on any factor. Goodness-of-fit of this solution was inadequate [χ2 (736) = 842.69; p =.004], suggesting that it should be rejected. When more than six factors were rotated, the additional factors had less than three substantial loadings and were not readily interpretable. Hence, the six-factor solution was preferable, supporting a distinct diabetes-specific dimension as part of the multidimensional structures of self-esteem.
DSSE scores were higher in boys than in girls, t(85) = 2.62, p =.01. Age showed significantly negative correlations with DSSE and with SEQ factors in the domains of school, body image, and sports/athletics (Table II). Age was also negatively correlated with global self-esteem (r = −.26, p =.02) and with adherence to the diabetes regimen (r = −.25, p =.02), and it was positively correlated with HbA1c (r =.25, p =.02), suggesting a general tendency for self-esteem and diabetes self-care indices to deteriorate with youth age. For that reason, age was entered first as a covariate in all subsequent hierarchical multiple regression analyses.
As expected, each of the domain-specific self-esteem factors evidenced highly significant zero-order correlations with ratings of global self-esteem (Table II). The hierarchical multiple regression results are shown in Table III. Age entered at step 1 explained 6%, and the five SEQ factors entered at step 2 together accounted for an additional 69% of the variance in global self-esteem. When entered at step 3, the DSSE factor uniquely explained an additional 2% of the variance [ΔF(1,79) = 6.21; p =.01]; it offered a small but significant contribution to ratings of global self-esteem. When this regression weight was constrained at zero, the resulting model showed an inadequate goodness-of-fit [χ2 (1) = 6.52; p =.01], indicating that the influence of DSSE needed to be included for the model to fit the data.
Next, we examined relationships between the self-esteem dimensions and diabetes self-care outcomes (i.e., adherence and glycemic control). DSSE and adherence as measured by the DSMP showed a highly significant correlation (r =.41, p <.001). SEQ factors pertaining to school, family, body image, and sports/athletics were also significantly associated with adherence (Table II). In hierarchical regression analysis predicting adherence, these SEQ factors together with age accounted for 18% of the variance (Table III). However, when DSSE was entered last into the equation, it uniquely explained a significant 4% variance increment in adherence [ΔF(1,79) = 4.05; p =.048]. In the full regression model, DSSE was the only predictor retaining a significant regression coefficient (β =.25, p =.048). Constraining this coefficient to zero resulted in a model with inadequate goodness-of-fit [χ2 (1) = 3.86; p =.050].
DSSE scores also evidenced a highly significant zero-order correlation with glycemic control (HbA1c) in the expected direction (r = −.36, p =.001). SEQ factors significantly related to HbA1c were school- and family-specific self-esteem (Table II). In hierarchical regression analysis, age and these SEQ factors in combination explained 11% of the variance in HbA1c (Table III). When DSSE was entered last, it significantly enhanced the proportion of variance explained by 5% [ΔF(1,73) = 4.25; p =.04]. In the full model, only the DSSE retained a significant multiple regression coefficient (β = −.29, p =.04), and constraining this coefficient to zero resulted in inadequate model fit [χ2 (1) = 4.43; p =.04].
To examine inter-rater reliability, youth self-esteem scores were compared with corresponding ratings obtained from their parents. Descriptive statistics for youth- and parent-report self-esteem forms are shown in Table IV. The mean scores of youths (M = 2.83) and parents (M = 2.80) were somewhat above the scale midpoint of 2.5; youths tended toward favorable self-evaluations in diabetes-specific contexts which was similarly reflected in their parents’ ratings. However, no participant had the highest possible DSSE score of 4.0, such that the ratings were not compromised by acquiescent response bias. Internal consistencies were adequate for both DSSE report forms (Cronbach α's of.87 for youth report and.80 for parent report, respectively). Youth- and parent-report DSSE scores did not significantly differ from one another, t(85) = 0.61, p =.54, and showed significant agreement, with an intraclass correlation coefficient (ICC) of.41 (p<.001). Agreement on DSSE was comparable to the level of agreement found between youth SEQ scores and parent CSEQ scores on corresponding subscales, with ICCs ranging from.21 (p =.03) to.45 (p <.001).
The development and maintenance of positive self-esteem is believed to be important for psychological adjustment to and successful management of a chronic illness during childhood and adolescence (Helgeson et al., 2007; Kovacs et al., 1992). Although self-esteem has been quite commonly assessed in youths with diabetes, understanding of its role and relevance is still vague and limited. Previous studies have relied on generic instruments developed with healthy youths and tapping areas of concern that are not directly targeted at experiences with the illness. This is a critical oversight, given that the demands and social expectations associated with diabetes constitute an enduringly salient life-context within which youths evaluate themselves, and within which they may be assumed to generate a specific sense of high or low self-esteem.
The primary objective of this study was to evaluate a new measure designed to assess a domain of self-esteem in adolescents with type 1 diabetes that directly pertains to their illness. The results provided promising preliminary evidence for the reliability and validity of this measure. The DSSE scores had high internal consistency and formed a unique dimension in factor analysis. Concurrent and incremental validity was supported in relation to youths’ global self-esteem and with measures of adherence and glycemic control. Agreement between self-report and parent-report forms of DSSE suggested inter-rater reliability of the scale.
Results from factor analysis replicated previous research indicating that adolescent self-esteem is multifaceted (DuBois et al., 1996; Marsh & Craven, 2006). Importantly, however, the factor solution supported the hypothesis that youths with diabetes develop a distinctive sense of self in relation to the demands posed by their illness, which cannot be sufficiently understood from how they view themselves in other areas of their lives. Furthermore, this diabetes-specific factor showed a small but unique contribution to ratings of global self-esteem beyond other domain-specific self-esteem factors in hierarchical regression analysis. This suggests that DSSE may be significantly involved in youths’ global feelings of worthiness as a person, which has important potential implications both theoretically and for clinical practice. Theoretically, it challenges the deficit-oriented perspective that has predominated research on self-esteem in youths with chronic illness, which presumes a uniformly negative impact of diabetes on youths’ feelings of self-worth (Hanson et al., 1990; Harper, 1991). Instead of imposing homogeneity, recognizing a diabetes-specific dimension may be vital for understanding individual differences in youths’ response to diabetes. For youths who develop a negative DSSE, this may adversely affect their overall sense of self; on the other hand, developing a favorable DSSE may just as well boost a youth's general feelings of self-worth. From a clinical standpoint, positive self-worth has been shown to be a vital aspect of general psychological well-being and to be protective of depression (Harter, 1999), such that supporting a favorable DSSE may be an important pathway to facilitating overall mental health in youths with diabetes.
Furthermore, from a clinical perspective, the finding that DSSE scores were higher among younger than among older youths deserves attention. A key concern in diabetes care is that, as adolescents get older, diabetes becomes more difficult and frustrating to control due to physiological and psychosocial changes (Mortensen et al., 1998). At the same time, family members have been found to decrease involvement in and support for the management of the illness during this developmental period (LaGreca & Bearman, 2002). Thus, it would not appear surprising that older youths have greater difficulty maintaining a positive sense of self in relation to their illness, which may be a risk factor for increased psychological distress.
In support of its concurrent and incremental validity, the DSSE scale showed highly significant correlations with adherence and glycemic control. These relationships remained significant when controlling for age in hierarchical multiple regression analyses, and DSSE explained adherence and glycemic control beyond traditionally studied dimensions of youth self-esteem. Pediatric investigators have emphasized that illness-specific measures may be especially helpful in understanding successful self-management and control of chronic disease (Drotar, 1997), and the results of this study are in accordance with this. Similarly, increasing empirical evidence suggests that measures of self-esteem and self-concept are more powerfully linked with performance and achievement outcomes when domain-specific measures in corresponding content areas are used (Swann, Chang-Schneider, & McClarty, 2007). So far, support for this claim has been mostly limited to healthy adolescent self-concept research. The present study demonstrates that recognizing this “specificity matching principle” (Swann et al., 2007, p. 87) is equally crucial for understanding the involvement of self-esteem in the self-management of type 1 diabetes in adolescence. In this regard, the DSSE scale may be useful to include in future studies of diabetes self-management in adolescence.
Finally, we found considerable agreement between youth self-reports and parent reports of DSSE, supporting the inter-rater reliability of the proposed measure. This result was encouraging in that it suggests that many parents can quite successfully infer how their children feel about themselves when it comes to managing their illness. However, it should also be noted that parent–child agreement, albeit sizeable, was far from perfect. Although we did not examine this, it might be speculated that conflicting views regarding DSSE may adversely affect successful disease management. In particular, those parents who are less able to recognize their child's internal experiences related to the illness may have greater difficulty providing adequate support for their child, and may be in greater need for intervention strategies targeted at family management of diabetes.
Although our initial findings with the DSSE scale were promising, continued study of this new scale is desirable and necessary. In particular, several study limitations suggest directions for further research. First, the limited sample size precluded the examination of the multidimensional self-esteem model with confirmatory factor analysis, and may be considered marginal for exploratory factor analysis. It should be noted, however, that the present analyses were built on a validated assessment that has exhibited well determined factors and consistently high communalities in previous studies (DuBois et al., 1996, 2000). Under these conditions, valid factor solutions “can be achieved with samples that would traditionally be considered too small for factor analytic studies, even when N is well below 100” (MacCallum, Widaman, Zhang, & Hong, 1999, p. 96). Nevertheless, the results of this study should be replicated with larger and more diverse samples. Second, the cross-sectional nature of the study restricts conclusions about causal relationships. In future research, longitudinal designs will be necessary to establish the causal order between DSSE, adherence, and glycemic control. We cannot say at this point whether DSSE is a cause or consequence of diabetes self-care outcomes. An intriguing additional possibility is a reciprocal effects model in which DSSE and self-care outcomes are mutually reinforcing, each leading to gains in the other (Marsh & Craven, 2006). Longitudinal studies will also be useful for examining developmental trajectories of DSSE over time. Third, HbA1c values indicating glycemic control were assessed at several different labs, rather than being assayed at a central lab and using a uniform laboratory method. This may have affected associations with the self-esteem measures by introducing error variance, thereby attenuating (i.e., weakening) the magnitude of correlations and potentially underestimating the actual size of effects. Finally, the present study did not provide insight into possible predictors and antecedents of DSSE. The identification of factors that facilitate the development and maintenance of a favorable DSSE over the course of adolescence is imperative, given the importance of these factors for the design of intervention strategies. In view of the theoretical perspective that self-esteem involves the comparison between one's actual self and one's desired self, perceptions of competence in managing diabetes should be linked to more favorable DSSE, and excessive or perfectionist standards for diabetes care should be linked with less favorable DSSE. In future studies it will be desirable to empirically test this theoretical prediction by examining whether DSSE is influenced differentially by measures of perceived competence (e.g., self-efficacy for diabetes management; Iannotti, Schneider et al., 2006) or factors that may increase competence (e.g., family support for diabetes; LaGreca & Bearman, 2002), and by factors that create excessive performance standards (e.g., excessive self-care autonomy; Wysocki et al., 1996).
In conclusion, the present study provided preliminary evidence for the usefulness of considering an illness-specific dimension of self-esteem in adolescents with type 1 diabetes, as well as for the reliability and validity of the proposed measure to assess this dimension. Pending further research, the DSSE scale may prove valuable for understanding how living with diabetes and the successful self-management of the illness are directly implicated in the development and maintenance of self-esteem. Although this study focused on a particular chronic illness, the concept of illness-specific self-esteem is not limited to type 1 diabetes. Future studies may wish to adapt the DSSE scale for other pediatric chronic conditions such as asthma, juvenile arthritis, or pediatric cancer, each of which creates a unique and salient context of experience within which youths evaluate themselves.
This work was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development.
Conflicts of interest: None declared.