Citation
Relationship among Family Functioning, Adherence, and Metabolic Control: Longitudinal Follow-Up

Material Information

Title:
Relationship among Family Functioning, Adherence, and Metabolic Control: Longitudinal Follow-Up
Creator:
GRABILL, KRISTEN M. ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Adolescents ( jstor )
Child psychology ( jstor )
Diabetes ( jstor )
Diabetes complications ( jstor )
Family structure ( jstor )
Insulin ( jstor )
Parents ( jstor )
Pediatrics ( jstor )
Regression analysis ( jstor )
Type 1 diabetes mellitus ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Kristen M. Grabill. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
5/31/2009

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

1 RELATIONSHIP AMONG FAMILY FUNC TIONING, ADHERENCE, AND METABOLIC CONTROL: LONGITUDINAL FOLLOW-UP By KRISTEN M. GRABILL A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Kristen M. Grabill

PAGE 3

3 TABLE OF CONTENTS page LIST OF TABLES................................................................................................................. ..........4 ABSTRACT....................................................................................................................... ..............5 CHAPTER 1 INTRODUCTION................................................................................................................... .7 Family Factors................................................................................................................. .........7 Adherence as a Mediator........................................................................................................ ..9 Intervention Research.......................................................................................................... ...10 Longitudinal Research.......................................................................................................... ..10 Study Goals.................................................................................................................... .........11 2 METHOD......................................................................................................................... ......12 Participants................................................................................................................... ..........12 Measures of Diabetes Specific Family Functioning...............................................................12 Measurement of Adherence....................................................................................................13 Measurement of Metabolic Control........................................................................................14 Procedure...................................................................................................................... ..........14 Data Analysis.................................................................................................................. ........14 3 RESULTS........................................................................................................................ .......15 Descriptive Statistics......................................................................................................... .....15 Mediation Analysis of Family Functioning, Adherence, and Two-Year HbA1c...................15 Mediation Analysis of Family Functioning, Adherence, and Change in HbA1c from Baseline to Two Years........................................................................................................17 4 DISCUSSION..................................................................................................................... ....22 LIST OF REFERENCES............................................................................................................. ..26 BIOGRAPHICAL SKETCH.........................................................................................................29

PAGE 4

4 LIST OF TABLES Table page 3-1 Intercorrelations between HbA1c, diab etes family measures, and adherence...................19 3-2 Regression analyses testi ng adherence as a mediator between family factors and HbA1c after two years.......................................................................................................20 3-3 Regression analyses testi ng adherence as a mediator between family factors and change in HbA1c across two years....................................................................................21

PAGE 5

5 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science RELATIONSHIP AMONG FAMILY FUNC TIONING, ADHERENCE, AND METABOLIC CONTROL: LONGITUDINAL FOLLOW-UP By Kristen M. Grabill May 2007 Chair: Gary R. Geffken Major: Psychology We examined the relationship among fa mily functioning, adherence, and metabolic control 2 years later in childre n and adolescents with type 1 diabetes, and examined the relationship among family functioning, adheren ce, and change in metabolic control from baseline to 2 years. Participants were 66 children (who partic ipated in a previous study) for whom psychosocial and metabolic control data were av ailable 2 years later. Measures of family functioning included the Diabetes Family Behavior Scale (DFBS), the Diabetes Family Behavior Checklist (DFBC), and the Diabetes Family Res ponsibility Questionnaire (DFRQ). Adherence was measured via the Diabetes Self-Management Profile (DSMP), a semi-structured interview. We collected HbA1c levels from medical records. Hierarchical linear regression was used to examine the paths between variables. Adherence partially mediated the relationshi p between family functioning variables and metabolic control after 2 years. Family functioning predicted a si gnificant amount of variance in 2-year change in metabolic control, but adherence did not mediate this relationship.

PAGE 6

6 A temporal relationship among family functi oning, adherence, and metabolic control was supported. Additionally, children who report the mo st negative family functioning may have the greatest decline in metabolic control. These findings have implicati ons for intervention and future research.

PAGE 7

7 CHAPTER 1 INTRODUCTION Type 1 Diabetes is an autoimmune disorder of the pancreas, affecting approximately 1 in 400 to 500 school aged children (American Diab etes Association, 2006). Onset peaks around age 10 to 12 years in girls, and 12 to 14 year s in boys (ADA, 2006). Children with poorly controlled diabetes are at risk for the short-term health compli cation of ketoacidosis, a condition that results in 10% of diabetes-re lated deaths under age 45 years. This subgroup is also at risk for long-term complications such as heart disease, kidney disease, blindness, and gangrene and indeed, two-thirds of individuals with diabetes will eventually die fr om heart disease (ADA, 2006). The Diabetes Control and Complications Trial (1993) demonstrated that risk for the above complications is reduced in those who c onsistently have good metabolic control. When individuals have blood glucose leve ls that are within a target range (ADA, 2006), they are said to be in good metabolic control. Unfortunately, metabolic control generally becomes worse over time (Jacobsen et al., 1994), highligh ting the need for intervention. Family Factors A large body of research has consistently de monstrated relationships between diabetesspecific family factors and metabolic control in children with type 1 diabetes. While studies examining associations between general family factors and metabolic control have shown mixed results or no relationship (G owers et al., 1995; Seiffge-Kre nke, 1998; Leonard, Jang, Savik, & Plumbo, 2005), diabetes-specific fa mily factors have shown stable relationships with metabolic control (McKelvey et al., 1993; Schafer, Glasgow , McCaul, & Dreher, 1983; Waller et al, 1986). For example, positive parental emotional suppor t (e.g., relating to the child about having diabetes) and parental guidance that is not perceived by the child as coercive are associated with improved metabolic control (Waller et al., 1986 ; McKelvey et al., 1993). One study used a

PAGE 8

8 behavioral observation measure to assess em otional support, sadness and anger surrounding diabetes diagnosis, and parent -child conflict resolution (Mar tin, Miller-Johnson, Kitzman, & Emery, 1998) and found that high levels of parental emotional support, ability to resolve sadness and anger surrounding diagnosis, and increased parent-child conflict resolution were all predictive of good metabolic control. Negative family factors have also been re lated to poor metabolic control outcomes. Patients in families reporting higher levels of conflict show relatively poorer metabolic control than those in families who report less conflict (Klemp & La Greca, 1987; Hauser et al., 1990; Miller-Johnson et al., 1994). Add itionally, parental behaviors spec ifically related to diabetes care that are perceived by the child as ne gative and unsupportive (e.g., coercion, threats, criticism, scolding, and nagging) are significantly related to poor metabolic control (Schafer et al., 1983; Schafer, McCaul, & Glasgow, 1986). Research also indicates th at disagreement between a parent and child about who is responsible for the implementation of the diab etes medical regimen is predictive of poor metabolic control (Anderson, Auslander, Jung, M iller, & Santiago, 1990). Anderson et al. (1990) reported that children with the poorest me tabolic control were from families in which neither parent nor child reported re sponsibility for diabetes-related tasks. In addition, Wysocki et al. (1996) found that child responsibility fo r diabetes management tasks was negatively associated with metabolic control. These investigations of individual diabetes -specific family factors have consistently revealed significant relationships between these variables and meta bolic control. However, those relationships have predicted a relatively small amount of va riance, and researchers have suggested that incorporating t hose variables into a larger model may yield stronger results

PAGE 9

9 (McKelvey et al., 1993; Lewin et al., 2006). Lewin et al. (200 6) investigated the following diabetes-specific family factors and their relations hip to metabolic control: parental warmth and caring, parental guidance and control, child pe rception of critical/uns upportive parents, and parent/child report that neither was responsible for the diabetes regimen. After controlling for demographic variables, these factor s together predicted 34% of the variance in metabolic control. This study highlights the importance of consider ing all family dimensions that relate to metabolic control, in order to obtain a complete model that may be translated into targeted intervention to improve metabolic control in children with type 1 diabetes. Adherence as a Mediator The medical management of type 1 diabet es is complex and can be challenging for children and adolescents due to the varied and in trusive nature of tasks required (e.g., monitoring of blood glucose, insulin injections , and dietary restraint). As ma ny as 45% of those with type 1 diabetes report some form of significant non-compliance during adolescence (Kovacs, Kass, Schnell, Goldston, & Marsh, 1989). For many youth, parents are involved in the administration of the diabetes regimen. As such, a mediati on model has been theorized in which diabetes specific family factors influence metabolic cont rol through their impact on adherence behaviors (Kovacs et al., 1989; Miller-J ohnson et al., 1994; Cohen, Luml ey, Naar-King, Partridge & Cakan, 2004; Lewin et al., 2006). In one study (Miller-Johnson et al., 1994) adherence fully mediated the relationship between parent-child conflict and metabolic c ontrol. In another, Cohen, Lumley, Naar-King, Partridg e, and Cakan (2004) tested adhe rence as a mediator between family dysfunction and metabolic control. While they did not find significant evidence for a mediating relationship, they al so did not examine diabetes-s pecific family functioning. Following these studies, Lewin et al. (2006) ex amined the mediating effects of adherence between combined family factors and metabolic control. Using path analysis, Lewin et al.

PAGE 10

10 (2006) demonstrated significant evidence for part ial mediation and established the relationship between family factors, adherence, and metabo lic control in a crosssectional design. While informative, the study was limited in that it did not establish a temporal relationship between variables. Intervention Research Further evidence for the role of family factor s comes from controlled trials of family psychotherapy for youth with type 1 diabetes. The majority of such studies have demonstrated improvement in psychosocial outcome (e.g., family c onflict, child behavior problems), but fail to demonstrate an improvement in metabolic contro l post-treatment (Harris, Harris, and Mertlich, 2005; Wysocki et al., 2000). Additionally, several studies have found improvements in metabolic control at post-trea tment, as well as psychosocial gains (Wysocki et al.,2006; Anderson, Brackett, Ho, and Laffel, 1999). Ho wever, although met hodologically rigorous treatment-outcome studies signify the importan ce of intervention for children with type 1 diabetes, results across studies have shown inc onsistent improvements in metabolic control. Those studies that have been successful in improving metabolic control do not delineate the mechanism by which change occurred. This inform ation is critical for understanding of the construct and for design of an optimal intervention. Longitudinal Research Most studies investigating the relationship between family factors and metabolic control have occurred within a cross-sectional design, li miting causal inferences that can be drawn. However, Jacobsen et al. (1994) followed a cohort of 61 children with type 1 diabetes across four years, and found that encouragement of expre ssiveness, greater family cohesion, and reduced conflict were associated with reduced degradatio n of metabolic control over time. However, no studies have investigated the combined impact of multiple family factors on metabolic control or

PAGE 11

11 adherence as a mediator of this relationship us ing a longitudinal design. In order to determine which elements are critical for designing inte rvention to improve metabolic control, the relationship between the above variable s requires temporal elucidation. Study Goals We used a longitudinal design to study the rela tionships among family factors, adherence, and metabolic control. Using a longitudinal de sign has two major advantages in testing the relationship between family factors and metabolic control. First, th e sequential order of variables in time allows greater causal inference than with a cross-sectiona l design. Second, the demonstrated tendency of metabolic control to worsen over time presents the possibility that family factors may have a cumulative effect; in other words, negative family factors, without intervention, may compound over time. Therefore, this study addresses the following aims: Aim 1: Test adherence as a mediator between a combination of diabetes specific family factors and metabolic control after two years. Aim 2: Test adherence as a mediator between diabetes specific family factors and change in metabolic control from baseline to two years.

PAGE 12

12 CHAPTER 2 METHOD Participants Participants were 66 children (52% male) with type 1 diabetes selected based on previous participation in Lewin et al. (2006; n = 108). All participants were required to have an HbA1c level two years after part icipation in that study. Participants ranged in age from 8 to 18 (M = 12.9 2.4) at the time of the initial study. The ethnicity of particip ants was: 78% Caucasian, 11% African American, 8% Hispanic, and 3% representing other ethnic groups. Participants were primarily from two parent households (77.3%). Participants in this study and those in the previous study did not di ffer significantly on any demographic variables, p s > .05. Measures of Diabetes Specific Family Functioning Diabetes Family Behavior Scale (DFBS). The DFBS (Walle r et al., 1986) is a childrated measure of perceived family support specific to diabetes. For this study, only the 15-item subscales measuring parental warmth and caring (e.g. “my parent understands how I feel about having diabetes”) and guidance a nd control (e.g. “my parent reminds me to test my blood sugar”) were used. Responses to this sc ale are given on a five-point s cale anchored by “all of the time” and “never”. The DFBS has show n good internal consistency, and test-retest reli ability (Waller et al., 1986; Lewin et al, 2006). Diabetes Family Behavior Checklist (DFBC). The DFBC (Schafer et al., 1986) is a measure of frequency of family support behaviors related to the diabetes regimen. For this study, the child-rated version was used. Additio nally, only the seven-item nonsupportive family behavior domain was used (e.g. “nag you about fo llowing your diet”). Responses are given on a five-point scale anchored by “never” and “at least once a da y”. The DFBC has shown good internal consistency (Schafer et al., 1986; Lewin et al., 2006).

PAGE 13

13 Diabetes Family Responsibility Questionnaire (DFRQ). The DFRQ (Anderson et al., 1990) is a measure completed separately by both the parent and child, and is meant to assess family sharing of responsibilit ies for diabetes management. This measure consists of 17 statements concerning diabetes management task s (e.g. remembering to take insulin), and asks the respondent to indicate which family member is responsible for that task (i.e. parent, child, or both). A “no-responsibility” score is calculated based on patterns of disagreement between the responses of the parent and that of the child. Hi gher scores indicate that neither the parent nor the child take responsibility for the task. Th e DFRQ demonstrates good internal consistency (Anderson et al., 1990; Lewin et al., 2006). Measurement of Adherence Diabetes Self-Management Profile (DSMP). The DSMP (Harris et al., 2000) is a 23item semi-structured interview. For purposes of this study, it was admini stered separately to both parent and child. Questions are designed to assess five areas of diabetes management: insulin administration and dose adjustment, blood glucose monitoring, exercise, diet, and management of hypoglycemia. Items are answered in an open-ended manner and coded by trained interviewers. Most items are scored on a five-point scale relative to the content of the question (e.g. “always eats more or gives more insulin”, “frequently eats more or gives less insulin”, “sometimes eats more or gives less insulin”, “occasionally eats more or gives less insulin”, “eats less than usual or gives more insulin”). Items from all domains are summed to produce a total adherence score, yi elding a separate parent score a nd child score. The DSMP has shown good internal consistency (Harris et al ., 2000; Lewin et al., 2006) and interobserver agreement (94%; Harris et al., 2000 ). Previous research with the DSMP has shown that it is strongly related to concurrent HbA1c (Lewin et al., 2005).

PAGE 14

14 Measurement of Metabolic Control Metabolic control is a biologica l assay of health status oper ationalized via the glycated hemoglobin A1c test (HbA1c). HbA1c provides an estimate of the glycemic control of a given individual over the previous 2-3 months ( ADA, 2003). Blood samples were analyzed using a Bayer DCA 2000+. Procedure Family functioning, adherence data, and base line HbA1c were obtained through archival records from participation in a previous study (L ewin et al., 2006). HbA1c levels were collected via finger-stick during routine medical visits and obtained for study purposes through the hospital’s computerized medical records inform ation system. All HbA1c levels occurred two years (+/two months) after participation in the previous study. Participants were consented for participation in the original study, and an IRB waiver of informed consent was obtained for collection of additional HbA1c data from archival medical records. Data Analysis Baron and Kenny’s (1986) guidelines for mediat ion were followed to test for mediating effects of adherence on the relationship between family functioning and long-term metabolic control. The following criteria are necessary for mediation: 1) the predictor (family functioning) should be significantly associated with the ou tcome (HbA1c), 2) the predictor should be significantly associated with the mediator (adherence), 3) the me diator should be significantly associated with the outcome (when the predicto r is accounted for), and 4) the addition of the mediator to the full model should reduce the rela tionship between the predictor and the criterion variable.

PAGE 15

15 CHAPTER 3 RESULTS Descriptive Statistics In order to control for demographic variab les in hierarchical regression, data were analyzed to test for relationships between demographic variables and family functioning, adherence, and HbA1c. Children from single parent families had significantly higher HbA1c levels at baseline (9.4%) and at two years (10.7 %) than children from two parent families at baseline (8.1%) and at two years (8.8%). However, family structure was not significantly related to any of the other family variables. Family structure was significantly related to child-rated adherence, t (65) = -48.54, p < .01. Family structure was also significantly related to baseline HbA1c, t (65) = 49.19, p < .01; and to two-year HbA1c, t (65) = -40.75, p < .01. All other demographic variables, including child’s age, duration with diabetes, and SES, were not significantly associated with a ny of the study variables. Inte rcorrelations among demographic and predictor variables are summarized in Table 3-1. Consistent with reported trends in HbA1c over time, HbA1c was significantly worse at two years ( M = 9.1%, SD = 1.7%) compared with baseline ( M = 8.3%, SD = 1.2%), t (64) = 5.27, p < .01. Additionally, participants who were included in the study by Lewin et al. (2006), but were not included in the present study due to unavailable HbA1c data ( n = 42), had significantly higher baseline HbA1c ( M = 9.2%, SD = 2.1) when compared to those in the present analysis ( M = 8.3%, SD = 1.3), t (107) = -2.42, p < .05. Mediation Analysis of Family Functi oning, Adherence, and Two-Year HbA1c Hierarchical linear regression was used to test each step of th e mediation model (see Table 3-2 for regression analyses related to each step of the guidelines for mediation). To control for demographic influence in all analyses, family structure (single or two parent family)

PAGE 16

16 was entered into block 1. The outcome variab le in each analysis was HbA1c unless otherwise indicated. Criterion 1. Family functioning variables (DFBS parental warmth and caring, DFBC critical and negative parenting, DFBS parental guidance and cont rol, and DFRQ parent-child dyad scores suggesting “no responsibility” for diab etes care) were entere d simultaneously into block 2. Results indicated that family variab les significantly predicte d 20% of variance in HbA1c two years later, F (2, 61) = 9.65, p < .01, providing evidence of a longitudinal relationship between family functioning and HbA1c. However, because DFBS parental guidance and control and DFRQ parent-child dyad scores suggesting “no responsibility” for diabetes care did not significantly contribute to the m odel at an alpha of .10, they were dropped from subsequent regression analyses. This appro ach, using an alpha of .10 for re tention of predictors in the regression analysis, is consistent with that used in similar regression models (Cohen et al., 2004; Davis et al., 2001). Criterion 2. With child-rated adherence as th e outcome variable, family functioning variables were entered into bl ock 2. Results indicated that family variables significantly predicted 23%, F (2, 61) = 10.56, p < .01, of the variance in child rated adherence, meeting the second requirement for mediation. Criterion 3. Child rated adherence pr edicted 7% of the variance in metabolic control with the effects of family factors accounted for, F (1, 60) = 7.25, p < .01, providing evidence for the third criterion. Criterion 4. Finally, when accounting for adhere nce in the model, the relationship between family factors and metabolic control was reduced from 20% to 8%, F (2, 60) = 4.06, p < .05,. The addition of the mediator reduced the si ze of the direct effect but did not reduce the

PAGE 17

17 effect to a non-significant valu e, suggesting partial mediation. As such, family factors were predictive of unique variance above and beyond adhere nce, indicating direct and indirect effects of family variables on metabolic control. Mediation Analysis of Family Functionin g, Adherence, and Change in HbA1c from Baseline to Two Years Hierarchical linear regression was used to test each step of the medi ation model (see Table 3-3 for regression analyses related to each step of the guidelines for mediation). To investigate changes in HbA1c over time controlling for base line HbA1c, baseline HbA1c was entered into block 1 of all analyses. To control for demogr aphic influences, family structure was always entered into block 2. The outcome variable in each analysis was HbA1c unless otherwise indicated. Criterion 1. Family functioning variables were ente red simultaneously into block 3 (DFBS parental warmth and caring and DFBC critical and negative parenting). Given that in the previous regression analyses, DFBS parental gui dance and control and DFRQ parent-child dyad scores suggesting “no responsibil ity” for diabetes care did not significantly contribute to the model, they were also excluded from this analysis. Results indicated that family variables significantly predicted 6% of variance in change in HbA1c across two years, F (2, 60) = 3.47, p < .05, providing evidence that family functioning predicts change in HbA1c over time. Criterion 2. Given that family functioning variable s and adherence were measured at the same time, and that adherence is the outcome me asure for this criterion, baseline HbA1c was not controlled for in this analysis. With child-ra ted adherence as the outcome variable, family functioning variables were entere d into block 2. Results indi cated that family variables significantly predicted 23%, F (2, 61) = 10.56, p < .01, of the variance in child rated adherence, meeting the second requirement for mediation.

PAGE 18

18 Criterion 3. Child rated adherence did not significantly pred ict variance in two-year change in metabolic control with the effects of family factors accounted for, F (1, 59) = 2.10, p = .15. Evidence was not provided for criterion 3 of Baron and Kenny’s (1986) steps for mediation. Therefore, step 4 was not completed and the me diating effect of adhe rence between family functioning and change in HbA1 c over time is not supported.

PAGE 19

19 Table 3-1. Intercorrelations between HbA1c, diabetes fam ily measures, and adherence. Variable 1 2 3 4 5 6 7 1 Baseline HbA1c ( M = 8.3% 1.2) ----.66** -.39** .50* -.08 .27** -.60** 2 Two-year HbA1c ( M = 9.1 1.7) -----.46** .29* -.18 .01 -.56** 3 Parental warmth and caring ( M = 53.4 9.0) -----.20 .21* -.06 .40** 4 Child report of critical and negative parenting ( M = 17.6 6.8) -----.03 .12 -.42** 5 Child report of parent al guidance and control ( M = 43.8 8.9) ----.10 .25* 6 “No responsibility” fo r diabetes care ( M = 2.47 2.6) -----.15 7 Child report of adherence ( M = 56.6 9.6) ----* p < .05. ** p < .01.

PAGE 20

20 Table 3-2. Regression analyses testing adherence as a mediator between family factors and HbA1c after two years. Block Variable(s) R2 R2 F VIF Criterion 1 1 .22 .22 18.02** Family structure* .42** 1.12 2 .41 .20 5.07** Child perception of parental warmth and caring -.40** 1.16 Child report of critical, unsupportive parents .15 1.10 Child report of parental guidance and control .02 1.14 “No responsibility” for diabetes care -.11 1.09 Criterion 2* 2 .35 .23 10.56** Child perception of parental warmth and caring .31** 1.05 Child report of critical, unsupportive parents -.31** 1.07 Criterion 3 2 .38 .20 9.65** Child perception of parental warmth and caring -.39** 1.05 Child report of critical, unsupportive parents .15 1.07 3 .45 .07 7.25** Child rated adherence -.32 1.53 Criterion 4 2 .37 .19 18.56 Child rated adherence -.46** 1.14 3 .45 .08 4.06* Child perception of parental warmth and caring -.29** 1.20 Child report of critical, unsupportive parents .05 1.22 *p < .05. ** p < .01. *Note: All regressions were conducte d with family structure controlled for in the first block. All regressi ons were conducted with HbA1c as the outcome variable, except the analysis in Criterion 2, in which adherenc e was the outcome variable. All standardized regression coefficients and multicollinearity statistic s are from the final block of each regression.

PAGE 21

21 Table 3-3. Regression analyses testing adherence as a mediator between family factors and change in HbA1c across two years. Block Variable(s) R2 R2 F VIF Criterion 1 1 .43 .43 48.27** Baseline HbA1c* .50** 1.69 2 .47 .04 4.50* Family structure* .22* 1.16 3 .53 .06 3.47* Child perception of parental warmth and caring -.25* 1.18 Child report of critical, unsupportive parents -.05 1.34 Criterion 3 3 .53 .06 3.47* Child perception of parental warmth and caring -.25* 1.18 Child report of critical, unsupportive parents -.05 1.34 4 .54 .02 2.10 Child rated adherence -.17 1.76 Criterion 4 3 .50 .03 3.51 Child rated adherence -.17 1.76 4 .54 .04 2.73 Child perception of parental warmth and caring -.22* 1.25 Child report of critical, unsupportive parents -.07 1.38 *p < .05. ** p < .01. Note: All regressions were conducte d with baseline HbA1c controlled for in the first block and family structure in the second block. Criterion 2 was excluded from this table because it is identical to Criterion 2 described in Table 2, in which baseline HbA1c was not entered into block 1 (see text). All standard ized regression coefficients and multicollinearity statistics are from the final block of each regression.

PAGE 22

22 CHAPTER 4 DISCUSSION This study examined the relationship between family functioning, adherence to a medical regimen, and long-term metabolic control in youth w ith type 1 diabetes. We tested adherence as a mediator of the relationship between two family factors (parental warmth and caring and child perception of unsupportive parent s) and metabolic control two years later. Results from regression analysis supported pa rtial mediation, in which family functioning impacted metabolic control both directly and via a dherence behavior. This is th e first study to demonstrate a temporal relationship between integrated family functioning variables, adherence, and metabolic control. Additionally, despite the two-year interval between the measurement of psychosocial variables and HbA1c, family functioning accounted for 20% of the variance in metabolic control. In a second set of analyses, th is study demonstrated that fa mily functioning significantly predicts a change in metabolic control over time. In other words, children reporting less parental warmth and caring and more parent al criticism at baseline are likely to deteriorate over time more quickly than children who report high levels of parental warmth and caring and low levels of parental criticism. This is particularly important given th at even a 1% increase in HbA1c (e.g., HbA1c = 8% at time one to 9% at time two) corresponds to a 15-30% increase in risk for microvascular and neuropathic complications of diabetes (ADA, 2003). This provides further evidence of the potential cumulativ e effect of suboptimal family functioning, and highlights the need for intervention in families dem onstrating maladaptive functioning. Our hypothesis regarding adherence as a medi ator of the relationship between family functioning and two-year change in metabolic control was not supported. There are several possible reasons that this relationship did not emer ge. First, adherence was measured at baseline, which was two years before the measurement of Hb A1c. This could be problematic given that

PAGE 23

23 adherence behaviors measured so far in adva nce would not influence HbA1c. Additionally, given evidence that adolescents are likely to experience periods of significant non-adherence (Kovacs et al., 1989), there is r eason to believe that adheren ce across childhood and adolescence is not a highly stable construct. This highlig hts the vulnerability of ch ildren and adolescents to changes in adherence behaviors and a declin e in HbA1c, providing further impetus for intervention during this critical period of development. The emphasis in this study was on adherence as a mediator of the relationship between family factors and metabolic control. However, it is likely that there is a reciprocal relationship between family functioning and adherence such that families become trapped in a coercive cycle. In other words, while parental criticism and lack of warmth may explain some non-adherent behavior, non-adherent behavior mi ght also elicit parent criticism. Over time, the frequency and intensity of parental negativity in creases, promoting further decrease s in child adherence. This is consistent with our findings, which demonstrat e that children reporting more negative family functioning show greater decreases in metabolic control, while those reporting more positive family functioning show the smaller decreases in metabolic control. These results have psychosocial treatment implications for improving metabolic control in children and adolescents with type 1 diabetes . Given that results of previous studies for intervention with these patients have been inconsiste nt, our results point to several critical factors that should be included to ma ximize success in future interventi ons. First, our results suggest that family functioning is related to metabolic control independently. Ps ychosocial interventions should target family functioning, with a particular emphasis on diabetes-specific areas of family functioning. Specifically, intervention should focus on improving parent-child communication and reducing factors that contri bute to family conflict (Geffken & Storch, 2006). Intervention

PAGE 24

24 should also focus on enhancing positive family fa ctors, such as warm and caring parent-child relationships, as well as concre te adherence behaviors (Heidgerk en et al., 2006). Therapists should acquire skills and knowledge specific to diabetes regimens to best facilitate family management of adherence (Adkins et al., 2006). Alarmingly, the results of this study also indicate that children having more negative family functi oning experience grea ter decline in metabolic control. Therefore, the need for efficacious intervention with these children is particularly critical. In order to intervene with at-r isk children, they must first be identified. When evaluating barriers to adherence, physicians should scre en for family functioning (Lewin et al., 2006). Additionally, given that this study identified singl e-parent families as a risk-factor for worse metabolic control, these families may warrant add itional attention. The measures used in this study can be used as screening tools in a clinical setting to identify at-r isk youth for referral to mental health services It is important to note the limitations of this study. First, family functioning variables and adherence were both measured at baseline, lim iting the causal inferences that can be drawn between the variables. In order to test the presence of a true r eciprocal relationship, they should both be measured at multiple times. Second, the study sample from which this sample was drawn included an additional 42 participants. Al though participants in that study were not aware that they would be participati ng in a later study (and thus canno t be labeled dropouts), there may be something systematically different about them that was not measured. Our analyses revealed a significant difference between those 42 participan ts and the participants in this sample, such that those who were not include d in the present analyses ha d significantly higher baseline HbA1c. It is possible that the present analyses excluded a less-healthy su bset of the population

PAGE 25

25 in need of study. However, our results repli cated the model found by Lewin et al. (2006), who used a sample comprised of all of these participants. Addition ally, it is conceivable that the reported relationship would be stronger in those who have worse metabolic control. Interestingly, our study found that two of the family factors did not have significant beta weights in the full model. While child perception of parental guidance and control, and report of “no responsibility” for the diabetes regimen were important predictors in the cross-sectional study (Lewin et al., 2006), they were not significant predictors in th is model. While this may be due to sampling differences, it may also indicate that these family factors have a greater shortterm impact, while the other family factors we measured (child perceptio n of warmth/caring and critical/negative pa renting) may have compounding effects on HbA1c over time. The present data also highlight ar eas in need of further explora tion. First, it is important to investigate the temporal relationship between fa mily functioning and adherence. This would help elucidate a reciprocal relationship between th e two variables. Similarly, further longitudinal study of family factors may determine whether th ese variables have a differential effect over time. Second, given the need for identification of children who are at-risk for the most rapid decline in metabolic control, there is a need fo r broad and well-recognized screening measures of family functioning. As such, a single brief que stionnaire based on family functioning variables that are predictive of metabolic control should be developed. Finally, in tervention studies based on this model may demonstrate improved unders tanding of the relationship between family functioning, adherence, a nd metabolic control.

PAGE 26

26 LIST OF REFERENCES Adkins, J., Storch, E.A., Lewin, A.B., Williams , L., Silverstein, J.H.. Malasanos, T.H, & Geffken, G. (2006). Home-based be havioral health intervention: Use of a telehealth model to address poor adherence to type-I diabetes medical regimens. Journal of Telemedicine and eHealth ., 12, 1-3. American Diabetes Association. (2003). Standards of medical care for patients with diabetes mellitus. Diabetes Care , 26 , S33–S50. American Diabetes Association. (200 6). Redefining diabetes control. Diabetes & Cardiovascular Disease Review, 1 , 1-8. Anderson, B. J., Auslander, W. F., Jung, K. C., Miller, J. P., & Santiago, J. V. (1990). Assessing family sharing of diab etes responsibilities. Journal of Pediatric Psychology , 15 , 477. Anderson, B. J., Ho, J., Brackett, J., & Laffel, L. M. B. (1999). An office-based intervention to maintain parent–adolescent teamwork in diabetes management: Impact on parent involvement, family conflict, and subsequent glycemic control. Diabetes Care , 22 , 713. Baron, R. M., & Kenny, D. A. (1986). The moderato r–mediator variable distinction in social psychological research: Con ceptual, strategic, and st atistical considerations. Journal of Personality and Social Psychology , 52 , 1173. Cohen, D. M., Lumley, M. A., Naar-King, S., Pa rtridge, T., & Cakan, N. (2004). Child behavior problems and family functioning as predictors of adherence and gl ycemic control in economically disadvantaged children with type 1 diabetes: A prospective study. Journal of Pediatric Psychology , 29 , 171. Davis, C. L., Delamater, A. M., Shaw. K. H., La Greca, A. M., Eidson, M. S., Perez-Rodriguez, J. E., & Nemery, R. (2001). Brief report: Pare nting styles, regimen adherence, and glycemic control in 4to 10-year ol d children with diabetes. Journal of Pediatric Psychology, 26 , 123129. Diabetes Control and Complications Trial (1993). The effect of in tensive treatment of diabetes on the development and progression of long-te rm complications in insulin-dependent diabetes mellitus. New England Journal of Medicine, 329 , 977-986. Geffken, G. & Storch, E. ( 2006). Psychological treatment of youth with Difficult-to-Manage Type 1 Diabetes. Current Trends in Endocrinology, 1, 127-135. Gowers, S. G., Jones, J. C., Kiana, S., Nort h, C. D., & Price (1995). Family functioning: A correlate of diabetic control? Journal of child Psychology and Psychiatry, 36 , 993-1001. Harris, M. A., Harris, B. S., & Mertlich, D. ( 2005). Brief report: In-home family therapy for adolescents with poorly controlle d diabetes: Failure to maintain benefits at 6-month followup. Journal of Pediatric Psychology, 30 , 683-688. Harris, M. A., Wysocki, T., Sadler, M., Wilkin son, K., Harvey, L. M., Buckloh, L. M., et al. (2000). Validation of a structured interview for the assessment of diabetes self-management. Diabetes Care , 23 , 1301. Hauser, S. T., Jacobson, A. M., Lavor i, P., Wolfsdorf, J. I., Herskowitz, R. D., Milley, J. E., et al. (1990). Adherence among children and adolescents with insulin-dependent diabetes mellitus over a four-year longitudinal fo llow-up. II. Immediate and long-te rm linkages with the family milieu. Journal of Pediatric Psychology , 15 , 527.

PAGE 27

27 Heidgerken, A., Storch, E.A., Williams, L., Lewin, A.B., Silverstein, J.H., Adkins, J., Malasanos, T., & Geffken, G.R. (2006).Telehealth interven tion for adolescents with type 1 diabetes . Journal of Pediatrics, 148, 707-708. Jacobson, A. M., Hauser, S. T., Lavori, P., Willett, J. B., Cole, C. F., Wolfsdorf, J. I., et al. (1994). Family environment and glycemic cont rol: A four-year prosp ective study of children and adolescents with insulin-dependent diabetes mellitus. Psychosomatic Medicine , 56 , 401– 409.Klemp & La Greca, 1987; Kovacs, M., Kass, R. E., Schnell, T. M., Goldst on, D., & Marsh, J. (1989). Family functioning and metabolic control of school aged children with IDDM. Diabetes Care , 12 , 409. Kovacs, M., Kass, R. E., Schnell, T. M., Goldst on, D., & Marsh, J. (1989). Family functioning and metabolic control of school aged children with IDDM. Diabetes Care , 12 , 409. Leonard, B. J., Jang, Y., Savik, K., & Plumbo, M. A. (2005). Adolescents with type 1 diabetes: Family functioning and metabolic control. Journal of Family Nursing, 11 , 102-121. Lewin, A. B., Heidgerken, A. D., Geffken, G. R., Williams, L. B., Storch, E. A., Gelfand, K. M., & Silverstein, J. H. (2006). Th e relation between family factor s and metabolic control: The role of diabetes adherence. Journal of Pediatric Psychology, 31 , 174-183. Lewin, A. B., Storch, E. A., Geff ken, G. R., Heidgerken, A. D., Williams, L. B., & Silverstein, J. H. (2005). Further examination of a structured adherence interview of diabetes for children, adolescents and parents. Children’s Health Care , 34 , 149-164. Martin, M. T., Miller-Johnson, S., Kitzman, K. M., & Emery, R. E. (1998). Parent-child relationships and insulin-dependent diabetes me llitus: Observational ratings of clinically relevant dimensions. Journal of Family Psychology, 12 , 102-111. McKelvey, J., Waller, D. A., North, A. J., Marks, J. F., Schreiner, B., Travis, L. B., et al. (1993). Reliability and validity of the di abetes family behavior scale. Diabetes Educator , 19 , 125– 132. Miller-Johnson, S., Emery, R. E., Marvin, R. S., Clarke, W., Lovinger, R., & Martin, M. (1994). Parent–child relationships and the management of insulin-dependent diabetes mellitus. Journal of Consulting and Clinical Psychology , 62 , 603. Schafer, L. C., Glasgow, R. E., McCaul, K. D., & Dreher, M. (1983). Adherence to IDDM regimens: Relationship to psychosocia l variables and metabolic control. Diabetes Care , 6 , 493. Schafer, L. C., McCaul, K. D., & Glasgow, R. E. (1986). Supportive and nonsupportive family behaviors: Relationships to adhe rence and metabolic control in persons with type 1 diabetes. Diabetes Care , 9 , 179. Seiffge-Krenke, I. (1998). The highly structured climate in families of adolescents with diabetes: Functional or dysfunctional for metabolic control? Journal of Pediatric Psychology, 23 , 313322. Waller, D. A., Chipman, J. J., Hardy, B. W., Hight ower, M. S., North, A. J., Williams, S. B., et al. (1986). Measuring diabetes-spe cific family support and its rela tion to metabolic control: A preliminary report. Journal of the American Academy of Child Psychiatry , 25 , 415. Wysocki, T., Harris, M. A., Buckloh, L. M., Mert lich, D., Sobel Lochrie, A., Taylor, A., Sadler, M., Mauras, N., & White, N. (2006). Effects of behavioral family systems therapy for diabetes on adolescents family relationships, tr eatment adherence, and metabolic control. Journal of Pediatric Psychology, 31 , 928-938.

PAGE 28

28 Wysocki, T., Linschied, T. R., Taylor, A., Yeates , K. O., Hough, B. S., & Naglieri, J. A. (1996). Deviation from developmentally appropriate self –care autonomy: Association with diabetes outcomes. Diabetes Care , 19 , 119.

PAGE 29

29 BIOGRAPHICAL SKETCH Kristen Grabill was born on October 3, 1981 in Washington, D.C. She grew up in Riva, MD, graduating from South Rive r High School in 1999. She graduated magna cum laude with a B.A. in psychology from Saint Mary’s Colle ge of Maryland in 2003. Following that, she completed an intramural post-baccalaureate fello wship at the National Institutes of Health in Bethesda, MD. Currently, Kristen is enrolled in the Clinical Psychology Ph.D. program at the University of Florida, and is expected to graduate in May 2010.