Maternal BMI is Independently Associated with Diabetes Control in Young Patients

Abstract

Objective

Investigate the cross-sectional association of glycemic control of ethnically diverse diabetic youth with family characteristics.

Design

Family study of 91 diabetic youth (probands) and 142 parents.

Results

Children’s age and HbA1c averaged 11.9 years and 8.9%, respectively; 69% were minorities. After adjustment, poor glycemic control was associated with minority race/ethnicity, more television viewing, and higher maternal body mass index (BMI). Average HbA1c was 1.2 and 1.9% units higher for children of overweight and obese mothers, respectively (p=0.004).

Conclusions

The positive association between maternal body composition and child HbA1c likely represents the unique behavioral influence of mothers.

Introduction

Improving glycemic control, as manifested by lowering levels of hemoglobin A1c (HbA1c), reduces the development of long-term complications of diabetes mellitus such as retinopathy, nephropathy, and neuropathy (The Diabetes Control and Complications Trial Research Group, 1993). Children with diabetes are especially vulnerable to develop these and other complications in early adulthood because of the duration of their disease (Donaghue, Chiarelli, Trotta, Allgrove, & Dahl-Jorgensen, 2009). Improving blood sugar management throughout childhood and adolescence, and therefore preventing or postponing the development of morbid complications, has immense patient, family, and societal implications.

Prior research investigating family-level risk factors for poor glycemic control and development of complications has been limited to genetic, clinical, and socioeconomic factors. For example, patients with diabetes who have siblings and/or parents with earlier onset of diabetic complications are more likely to have microvascular complications (Monti et al., 2007). Additionally, worse glycemic control and increased hospitalization rates are seen among patients from lower socioeconomic strata, minority groups, and underinsured families (Palta et al., 1997)(Swift, Chen, Hershberger, & Holmes, 2006).

Microvascular complications of diabetes, such as nephropathy and retinopathy, also aggregate within families, and several candidate genes have been identified (Abhary, Hewitt, Burdon, & Craig, 2009)(Boright et al., 2005)(Pezzolesi et al., 2009). Some evidence suggests increased genetic risk for coronary heart disease among persons with type 2 diabetes (Ruiz et al.,1994). Without question, genes play an essential role in developing complications; however, family environment also impacts an individual’s ability to control glycemia and thereby prevent complications.

Research specifically on the health habits of family members, including parental body habitus, perceived stress, and exercise, eating, and television practices, has not been conducted with reference to families of children with diabetes. Adverse parental habits may place their children at greater risk of poor diabetes management. We addressed this question by investigating the association of glycemic control with these family characteristics in an ethnically and socioeconomically diverse group of families of youth with diabetes from the Chicago metropolitan area.

Methods

Study Sample

The Chicago Childhood Diabetes Registry Family Study traced the epidemiology and natural history of childhood diabetes in an ethnically diverse community sample. We recruited through diabetes clinics, health fairs, and mailings and invited the index cases, children with any type of diabetes, to participate if they were <18 years of age at diagnosis and diabetes was not secondary to another condition. All biological first- and second-degree relatives were also invited to participate, however non-parental/guardian participant data were not included in this analysis. We examined participants in their homes or in the General Clinical Research Center at the University of Chicago. Institutional Review Board approval was obtained. Parents provided informed consent and children 10–17 years old gave assent. This analysis focuses on the 91 patients who were <18 years of age at the time of their participation, and includes information on their parents/guardians (n=142) and family environment.

Data Collection

Data were collected via interviews, onset medical records, physical examinations, and biospecimens. Individual interviews collected information related to demographics, diabetes treatment and management, and health practices. In addition, the families selected one informant to provide family-level demographic and behavioral data.

Demographics

We defined patient race/ethnicity as that reported for three or more grandparents; if <3 grandparents shared the same race/ethnicity, the child was considered of mixed origin. When race/ethnicity was only available on one or two grandparents, parental data were used. HbA1c outcomes for non-Hispanic Black, Hispanic, and other/mixed families did not differ significantly and the three groups were therefore combined and classified as minority race/ethnicity. Heads of household (defined as parent, main wage earner, or the person who makes the important decisions) reported health insurance status and total household income. Insurance was categorized into private insurance vs. Medicaid/no insurance, and household income was dichotomized into below or at least $50,000 per year. Mothers reported their highest level of education, which was dichotomized into school beyond high school vs. high school or less. Dichotomized income level was imputed for four missing responses using a prediction model based on food insecurity, number of parents in the home, maternal education level, and health insurance.

Diabetes Characteristics

Parents who either reported having diabetes or had an elevated fasting glucose (≥126 mg/dL) measured by a glucometer (One Touch Sure Step, Lifescan, Milpitas, CA) were characterized as having diabetes in this analysis. Patients were definitively classified as having Type 1 diabetes based on undetectable C-peptide, or detectable C-peptide with <2 years duration and positive islet autoantibodies (GAD, insulinoma-associated protein 2) (n=80). The other patients were defined as having non-type 1 diabetes (n=11; four of these did not use insulin). Fasting plasma C-peptide was measured in all patients. Those with a fasting blood glucose <150 mg/dL also had a stimulated plasma C-peptide measurement 90 minutes after ingestion of a 6 ml/kg standard nutrient solution (Boost, Novartis Nutrition Corporation, Minneapolis, MN). C-peptide was determined with a solid-phase, competitive chemiluminescent enzyme immunoassay (Immulite 2000, Diagnostic Products Corporation, Germany) in the University of Chicago’s Diabetes Research and Training Center Lab. The lower limit of detection was 0.17 nmol/L and the intra-assay coefficient of variation (CV) was 8%. Detected C-peptide in patients whose fasting glucose was >=150 mg/dL was considered stimulated. Absent C-peptide was defined as level below the detection limit. Antibodies to radiolabelled recombinant human GAD 65 (whole) and human insulinoma-associated protein 2 (349 AA cytoplasmic portion) were quantified by fluid-phase immunoprecipitation assay. Children (or their guardians) were asked questions regarding their treatment regimen (insulin, pills, both, or diet). We calculated age at diagnosis and duration with diabetes based on information from the onset medical record; if not available (n=6), date of diagnosis was self-reported. HbA1c in whole blood was measured with the DCA 2000+ Analyzer (Bayer Healthcare, Elkhart, IN) using a latex immunoagglutination inhibition method. The intra- and inter-assay coefficients of variation were <4.3%, and the detection range for HbA1c was 2.5–14.0% (27–153 mmol/mol).

Family Environment

Single vs. two-parent household was determined for each family. Heads of household were asked the six-item version of the USDA Household Food Security Scale, designed to measure a household’s ability to access enough food to fully meet their basic needs, for example: “In the last 12 months, did you ever cut the size of your meals because there wasn’t enough money for food?” (Blumberg, Bialostosky, Hamilton, & Briefel, 1999). Families were considered food insecure if they responded affirmatively to two or more items. All participants were asked how often they eat together with people in their household and how many hours per week they spend viewing television. Parents answered the four-item version of the Perceived Stress Scale, a validated measure of the degree to which situations in one’s life are appraised as stressful, for example: “In the last month, how often have you felt unable to control important things in your life?” (Cohen, Kamarck, & Mermelstein, 1983). The possible range of scores was 0–16, where a higher value indicates higher stress levels. Additionally, parents assessed their family’s support for exercise using a 13-item Support for Exercise Scale (Sallis, Grossman, Pinski, Patterson, & Nader, 1987). These questions asked respondents to rate, for example, how often during the past 3 months their family “Exercised with [him/her],” or “Gave [him/her] encouragement to stick with an exercise program.” Higher values indicated greater support, with a possible range of total scores being 13–65. Parents also reported their exercise frequency, dichotomized for analysis as exercising at least one day per week vs. less.

Anthropometrics

At the time of the interview, using standardized methods (Lohman, Roche, & Martorell, 1988), height was measured without shoes using a stadiometer rod, and weight (all participants) and percent body fat (those >9 years old and not pregnant) were measured with a bio-electrical impedance analyzer scale (Tanita TBF-300A, Arlington Heights, IL). Sixty-three of the patients were >9 years old and able to have valid body fat measurements. Body mass index (BMI) of children was transformed into Z-scores using age- and sex-matched reference data from the 2000 Center for Disease Control Growth Charts. Maternal and paternal BMI was categorized into normal (<25kg/m²), overweight (25–30 kg/m²), and obese (>30kg/m²). BMI was not calculated for two pregnant mothers.

Insulin Resistance

Serum insulin was measured with a solid-phase, two-site chemiluminescent immunometric assay (Immulite 1000, Siemens Medical Solutions Diagnostics, Los Angeles, CA) by the University of Chicago’s Diabetes Research and Training Center Laboratory. Using fasting insulin and glucose, insulin resistance was determined for all participating parents without type 1 diabetes (n=133) using the Homeostasis Model Assessment version 2.0 (Matthews et al., 1985) (Wallace, Levy, & Matthews, 2004), available online (The Oxford Centre for Diabetes, 2007). This mathematical model is a widely used tool to estimate insulin sensitivity and β–cell function. Insulin resistance was not calculated for two pregnant mothers.

Statistical Methods

Linear regression analyses were conducted to identify significant correlates of the dependent variable, child’s HbA1c. The association of each covariate with child HbA1c was evaluated individually; then covariates with p-value less than 0.15 were entered into multivariable analysis. Separate multivariable models were fit for each of the covariate categories: patient characteristics, family characteristics, and maternal characteristics. Covariates significant in these models were combined into an overall multivariable model, and then non-significant covariates were dropped using a backwards selection approach to obtain a parsimonious model (Chatterjee & Hadi, 2002). Covariates not included in the parsimonious model were checked individually to determine whether they confounded the association between patient HbA1c and any of the covariates in the model. Confounding was considered present if any regression coefficient changed by 10% or more when the potential confounder was added back into the model. No confounders were identified. Patient age, patient BMI Z-score, and maternal diabetes were added to the parsimonious model to obtain the final multivariable model; these covariates were included due to their historical association with HbA1c in previous studies and to facilitate comparison with analyses in similar studies. This multi-step approach to building the multivariable regression model was taken due to limitations on the recommended number of covariates for the given sample size. All analyses were conducted using STATA version 10.0 for Macintosh (College Station, TX).

Four sensitivity analyses were conducted by excluding certain subgroups from the final multivariable model (data not shown). In the first, patients not using insulin (n=4) were excluded; in the second, patients classified as having non-type 1 diabetes (n=11) were excluded. Neither of these exclusions substantially changed the results; all children were therefore included in the final analysis. A third sensitivity analysis excluded the two guardians who were aunts, again with no substantial change in the results; they were considered mothers for the purpose of this analysis. Finally, due to the higher number of mothers providing body composition data (n=89) as compared to fathers (n=51), we included only those families with two participating parents. In this restricted subgroup, we found results similar to the original model.

Results

Participants in the current analysis (Table 1) included 91 children with diabetes, 89 biological mothers, 51 biological fathers, and 2 biological aunts/guardians. Participants were racially and ethnically diverse, with 31% of the patients Non-Hispanic White, 42% Non-Hispanic Black, 17% Hispanic, and 10% Other/Mixed race. For the children, mean age at examination was 11.9 years, ranging from 2.6 to 17.9 years. The majority used insulin monotherapy to treat their diabetes; however, three used pills only, two a combination of insulin and pills, and one diet alone. Eighty (88%) were definitively classified as having Type 1 diabetes. Mean(standard deviation) HbA1c was 8.9(2.3)% or 74(25) mmol/mol, and ranged from 4 to 14% or 20–130 mmol/mol; mean BMI Z-score was −0.8 (0.7). For the parents, ten reported having diabetes; a further 5 parents had fasting glucose results within a diabetic range. Mean BMI for mothers and fathers was 30.0 and 29.3 kg/m², respectively. Forty-eight percent of the households had Medicaid or no health insurance, 40% were single parent homes, and 20% were considered food insecure. The patients watched 20 hours of television per week, on average.

Table 1.

Descriptive Statistics of Patients and Their Parents

Characteristic	N	N (%)	Mean (Standard Deviation)	(Range)
Demographics
Age (years)	91		11.9 (4.0)	(2.6 –17.9)
Male Sex	91	49 (54)
Minority Race	91	63 (69)
Family Income <$50,000/year	91	50 (55)
Medicaid/No Insurance	91	44 (48)
Maternal Education (High School or less)	91	26 (29)
Child’s Diabetes Characteristics
Age at Diagnosis (years)	91		7.7 (4.0)	(0.9–15.9)
Duration with Diabetes (years)	91		4.3 (3.2)	(0.1–15.6)
HbA1c (%); mmol/mol	91		8.9 (2.3); 97.3 (25.1)	(4.0–14.0);(44–153)
Type 1 Diabetes	91	80 (88)
Family Environment
Single Parent Home	91	36 (40)
Food Insecurea	91	18 (20)
Family Eating Habits (eats together everyday)	90	42 (47)
Child’s TV Viewing (hours/week)	90		20.0 (17.6)	(2–105)
Maternal TV Viewing (hours/week)	90		17.3 (15.6)	(0–119)
Paternal TV Viewing (hours/week)	50		12.0 (7.5)	(1–39)
Maternal Perceived Stress b	91		6.0 (2.7)	(1–15)
Maternal Family Support for Exercise c	76		32.2 (9.3)	(19–59)
Maternal Exercise (at least one day/week)	90	64 (71)
Paternal Perceived Stress b	50		4.3 (2.9)	(0–11)
Paternal Family Support for Exercise c	43		31.5 (9.6)	(21–53)
Paternal Exercise (at least one day/week)	50	33 (66)
Child’s Body Composition
BMI Z-score	91		−0.8 (0.7)	(−2.0–1.2)
Body Fat (%)	63		27.4 (10.0)	(5.2– 45.9)
Maternal Body Composition
BMI (kg/m2)	88		30.0 (7.4)	(17.5–48.3)
Body Fat (%)	87		39.5 (8.3)	(20–54.8)
HOMA-IR	87		1.4 (2.1)	(0.3–18.2)
Paternal Body Composition
BMI (kg/m2)	51		29.3 (5.6)	(15.8–43.6)
Body Fat (%)	51		27.1 (7.6)	(8.4–43.2)
HOMA-IR	46		1.2 (0.9)	(0.3–5.5)
Parents with Diabetes
Mother with diabetes	89	10 (11)
Father with diabetes	51	5 (10)
Either parent with diabetes	91	15 (16)

HbA1c = Hemoglobin A1c; BMI = Body-mass index; HOMA-IR = Homeostatic Model Assessment- Insulin Resistance;

^aHousehold Food Security Scale, insecure if answered affirmatively to at least 2 of 6 questions;

^b4-item version of Perceived Stress Scale, higher value= greater stress, range 0–16;

^c13-item Support for Exercise Scale, higher value= greater support, range 13–65.

In univariate analysis (Table 2a), several demographic factors were significantly associated with poor glycemic control in the children: older age at exam, minority race, lower family income, and Medicaid/no insurance. In addition, the association between older age at diagnosis and higher HbA1c approached statistical significance (p=0.09). Duration of diabetes was not significantly associated with HbA1c.

Table 2.

Table 2a. Univariate Associations of HbA1c(%) with Patient, Parent, and Family Characteristics
Characteristic	N	β-coefficient % (mmol/mol)	(95% C.I.-%) (95% C.I.- mmol/mol)	Intercepta % (mmol/mol)	P-value
Demographics
Age (years)	91	0.14 (1.53)	(0.02, 0.26) (0.22, 2.84)	8.9 (73.8)	0.018
Minority Race vs. Non-Hispanic White	91	1.63 (17.8)	(0.66, 2.60) (7.21, 28.4)	7.8 (61.7)	0.001
Family Income <$50,000/year vs. More	91	1.62 (17.7)	(0.72, 2.51) (7.87, 27.4)	8.0 (63.8)	0.001
Medicaid/No Insurance vs. Private	91	1.11 (12.1)	(0.19, 2.03) (2.08, 22.2)	8.4 (68.3)	0.019
Age at Diagnosis (years)	91	0.10 (1.1)	(−0.02, 0.22) (−0.22, 2.40)	8.9 (73.8)	0.09
Family Environment
Single Parent Home vs. Two	91	1.30 (14.2)	(0.37, 2.23) (4.04, 24.4)	8.4 (68.3)	0.007
Child’s TV Viewing (hours/week)	90	0.04 (0.44)	(0.02, 0.07) (0.22, 0.77)	8.9 (73.8)	0.001
Food Insecure	91	0.48 (5.24)	(−0.71, 1.66) (−7.76, 18.1)	8.8 (72.7)	0.43
Family Eating Habits (eats together every day)	90	−0.22 (2.40)	(−1.18, 0.74) (−12.9, 8.09)	9.0 (74.8)	0.65
Child’s Body Composition
BMI Z-score	91	0.31 (3.39)	(−0.36,0.97) (−3.93, 10.6)	9.1 (76.0)	0.36
Body Fat (%)	63	−0.02 (0.22)	(−0.08, 0.05) (−0.87, 0.55)	9.2 (77.0)	0.58
Maternal Body Composition
BMI Categories (kg/m2)	88				<0.001
Normal Weight (<25)	26	Reference		7.4 (53.4)
Overweight (25–30)	24	1.66 (18.1)	(0.50, 2.82) (5.46, 30.8)		0.005
Obese (>30)	38	2.46 (26.9)	(1.42, 3.51) (15.5, 38.4)		<0.001
Body Fat (%)	87	0.10 (1.1)	(0.04, 0.16) 0.44, 1.75)	8.9 (73.8)	0.001
HOMA-IR	87	0.34 (3.72)	(0.12, 0.57) (1.31, 6.23)	8.9 (73.8)	0.004
Paternal Body Composition
BMI Categories (kg/m2)	51				0.30
Normal Weight (<25)	9	Reference		7.8 (61.7)
Overweight (25–30)	23	0.51 (5.57)	(−1.10, 2.12) (−12.0, 23.2)		0.53
Obese (>30)	19	1.21 (13.2)	(−0.45, 2.87) (−4.92, 31.4)		0.15
Body Fat (%)	51	0.01 (0.11)	(−0.07, 0.09) (−0.77, 0.98)	8.5 (69.4)	0.81
HOMA-IR	46	0.18 (1.97)	(−0.49, 0.85) (−5.36, 9.29)	8.5 (69.4)	0.59
Parents with Diabetes
Mother with diabetes	89	1.33 (14.5)	(−0.17, 2.83) (−1.86, 30.9)	8.8	0.08
Father with diabetes	51	0.94 (10.3)	(−1.00, 2.88) (−10.9, 31.5)	8.4	0.33
Either parent with diabetes	91	1.14 (12.5)	(−0.12, 2.40) (−1.31, 26.2)	8.7	0.08

Table 2b. Variables Significantly Associated with HbA1c (%) in Multivariable Modelb (n=88)
Characteristic	β-coefficient %(mmol/mol)	(95% C.I.-%) (95% C.I.- mmol/mol)	P-value	Intercepta %(mmol/mol)
Minority Race vs. Non-Hispanic White	1.19 (13.0)	(0.15, 2.23) (1.64, 24.4)	0.026	6.0 (42.1)
Child’s TV Viewing (hours/week)	0.02 (0.22)	(0.00001, 0.05) (0.0001, 0.55)	0.05
Maternal BMI Categories (kg/m2)			0.004
Normal weight (<25)	Reference
Overweight (25–30)	1.24 (13.6)	(0.13, 2.36) (1.42, 25.8)	0.03
Obese (>30)	1.92 (21.0)	(0.81, 3.02) (8.86, 33.0)	0.001

CI=confidence interval

^aCentered at sample mean for continuous variables

^bAdjusted for patient age, family income, insurance status, patient BMI Z-score, and maternal diabetes. Only variables with p<0.05 shown. Adjusted R² = 0.31

We also explored various characteristics of the family environment. Children from single-parent homes were significantly more likely to have worse glycemic control. Additionally, there was a significant positive association between hours of television watched by the children and HbA1c. Food insecurity and family eating habits were not associated with the glycemic control of patients. Maternal and paternal perceived stress, perceived family support for exercise, and parental exercise and television habits were also not significantly associated with the child’s HbA1c (data not shown).

Patient, maternal, and paternal obesity measures were analyzed for relationships with child’s HbA1c. A child’s glycemic control was not significantly associated with his/her BMI Z-score, his/her percent body fat, paternal body BMI, paternal body fat, or paternal insulin resistance. In contrast, worse patient glycemic control was associated with greater maternal BMI, maternal percent body fat, and maternal insulin resistance. On average, in unadjusted analysis, overweight and obese mothers had children with HbA1c values 1.7 and 2.5 %-units (18.6 and 27.3 mmol/mol) higher than children of normal weight mothers, respectively (p= <0.001).

The presence of a diabetic parent in the family showed a trend toward significant correlation with worse glycemic control of the child (p=0.08); the magnitude of the association was larger for maternal diabetes than for paternal diabetes.

In multiple regression analysis (Table 2b), higher patient HbA1c was significantly and independently associated with minority race, more hours of television watched per week by the child, and higher maternal BMI, after controlling for child’s age at exam and BMI Z-score, family income, insurance status, and maternal diabetes. On average, patient HbA1c was 1.2 %units (13.1 mmol/mol units) higher among those of minority race (p=0.03), 0.02 %-units (0.22 mmol/mol units) higher for each hour of TV watched per week (p=0.05), and 1.2 and 1.9 %-units (13.1 and 20.8 mmol/mol-units) higher for children of overweight and obese mothers, respectively (p=0.004). These independent variables explained 31% of the total variance in patient HbA1c. First order interactions of maternal BMI with child BMI Z-score, age at exam, and gender on HbA1c were tested and none were significant.

Family income attenuated the effect of race while insurance status had the opposite confounding effect; however, when both variables were included in the model, these opposing effects cancelled and no confounding occurred. Maternal body fat and maternal BMI are measuring the same construct of maternal adiposity. When maternal body fat replaced maternal BMI in the final models, there was not a significant change to the strength of the associations of other variables with HbA1c. Therefore, maternal BMI was used in the final model due to ease of interpretation and clinical application.

Discussion

In this analysis of 91 families of children with diabetes, we examined a number of social, economic, and behavioral factors within the family. While prior research on children with diabetes includes the child and his/her accompanying parent, this study is unique because of its broader family focus and the participation of many minorities. We found several indicators of economic status to be highly correlated with elevated HbA1c, including low income, minority race/ethnicity, Medicaid or no insurance, and single parent homes. Food insecurity, which correlates with higher BMI in other studies (Casey et al., 2006), and families frequently eating together, were not associated with glycemic control in this analysis. Our study found a strong association of greater maternal body habitus with higher HbA1c of young patients with diabetes. This finding is independent of race/ethnicity, family income, insurance status, child’s age, child’s body habitus, hours watching television, and maternal diabetes. Of note, although the average BMI of parents in this sample is high, it is consistent with estimates of national averages for low-income and minority families (Wang & Beydoun, 2007). Additionally, though there were fewer participating fathers, the variability of paternal body composition data was similar to that of mothers; therefore restricted variability cannot account for the lack of association with patient HbA1c.

To our knowledge, the relationship between a mother’s body composition and her child’s glycemic control has not been described in the existing literature. Many studies have examined the relationship of maternal weight to diabetes onset; however, these studies typically focus on pre-pregnancy weight and its relationship with offspring adiposity and/or diabetes incidence (Boney, Verma, Tucker, & Vohr, 2005). Additionally, maternal type 2 diabetes, for which obesity is a major risk factor, correlates with worse glycemic control (Gong, Kao, Brancati, Batts-Turner, & Gary, 2008) and diabetic nephropathy (Hadjadj et al., 2007) in their offspring. Our results are the first to suggest that maternal body composition may have a significant impact on the glycemic control of offspring with diabetes. Perhaps the association between mothers’ weight and their child’s glycemic control could be explained by evidence that maternal BMI is positively associated with insulin resistance in non-diabetic offspring (Kazumi, Kawaguchi, & Yoshino, 2005). The young patients with diabetes in the current study who had heavier mothers may be more insulin resistant than their counterparts with leaner mothers, and therefore have a more difficult time maintaining glycemic control; unfortunately, there is not a simple way to measure insulin resistance in individuals with type 1 diabetes. Additionally, our study suggests that the association of maternal BMI with a child’s glycemic control is independent of family tendency toward diabetes as adjustment for either maternal diabetes status or maternal insulin resistance did not change the magnitude of the association.

The mechanism of this relationship must therefore be further investigated. It is generally acknowledged that there is a substantial genetic influence on body composition, yet in our data paternal body habitus was not related to patient HbA1c despite equivalent genetic contribution and an adiposity distribution that was similar to that of the mothers. This relationship continued to hold when we looked only at those families with two participating parents. Therefore, the strength of the association between glycemic control and mothers’ adiposity, independent of paternal variables, likely represents the unique behavioral and environmental influence of mothers on the glycemic control of their children with diabetes. In the clinical setting, this may warrant particular attention to families of diabetic children with overweight mothers. Nurses might use maternal adiposity as one indication that specific nursing interventions to improve lifestyle choices should be explored. Mothers are often the primary caretakers, including management of insulin and dietary intake of their children. Additionally, their own exercise habits may be passively or actively modeled for their diabetic children.

Our analysis includes several related socioeconomic factors whose impacts cannot be fully dissociated statistically from each other or from minority race. For example, the robust association of minority race with HbA1c of the patient, even after adjusting for family income and insurance, could be attributed to the residual socioeconomic impact of race in this sample.

There are limitations to our study. Recruitment was non-random and may represent a bias toward families that were interested in the study and proactive enough in their care to present to clinic, attend diabetes fairs, or respond to mailings. However, the poor glycemic control seen in our study population does not support a bias toward adherent families with well-controlled diabetes. Additionally, HbA1c is certainly a traditional indicator of the adequacy of diabetes management, but it represents only one aspect of this multifaceted construct and in this study was measured at just one time point. The data are cross-sectional, yet the impact of family traits on a young person’s disease outcomes will clearly be greater over the course of many years. We did not have longitudinal information on mothers’ body habitus, although based on the current literature it might be hypothesized that long-term maternal obesity could be one crucial determinant of glycemia in diabetic offspring. Self-reported behaviors, as included here, typically contain substantial error due to under- or over-reporting and social desirability biases within the family-centered interview, which cannot easily be accounted for. Finally, the sample of 91 patients and their families is not large enough to allow full exploration of all of the potential associations or to exclude the possibility of type II error. Of particular concern in this regard is the relatively small number of fathers who participated.

Existing research on children with chronic diseases suggests several possible mechanisms for the impact of family relationships on disease management which have important implications for nursing. Family climate may impact patients’ physiology directly by diet and exercise habits. Mothers may influence their children by modeling concern and behavior toward improving their own health issues. Some data exist showing that children who perceive their mothers as collaborators in their diabetes management have better adherence to clinicians’ recommendations and improved metabolic control (Wiebe et al., 2005). Our findings suggest that adult health behaviors may significantly impact child health outcomes. Maternal BMI may represent a combination of behavioral factors such as eating behaviors, exercise habits, self-regulation capacity, and health self-efficacy and attitude, which is modeled for their children. Poorer glycemic control of their children may be a manifestation of adverse aspects of the home environment. The significant and independent association of television viewing with glycemic control of the patient highlights the behavioral and attitudinal aspects of diabetes management. Children who spent more time in sedentary activities or who had mothers with poor self-regulation were more likely to demonstrate worse glycemic outcomes. Thus, nurses should consider devoting particular attention to families where the parents appear to have their own weight issues. This analysis has found that glycemic control of children with diabetes is strongly associated with their mothers’ body habitus. For nurses, this result emphasizes the need to take family practices into consideration with young patients with diabetes. The optimal holistic approach to treating these vulnerable children must include not only patient-focused but family-focused approaches to lifestyle changes. There is some evidence that family interventions are effective in improving glycemic control, particularly when the interventions target diabetes-related stress and family conflict (Fisher & Weihs, 2000). Future nursing research among children with chronic illnesses should consider family health care attitudes and behaviors. Altering the caretakers’ health beliefs and lifestyles may have both direct and indirect influences on the health of their children with diabetes.

Abbreviations

HbA1c

Hemoglobin A1c

BMI

body mass index

HOMA-IR

Homeostatic Model Assessment-Insulin Resistance