Although there remain questions about whether DEB is associated with poorer long-term metabolic control in patients with T1D (Affenito 1997a, Herpertz 1998a, Engstrom 1999, Peveler 2005), the presence of diagnosable ED and behavior categorized as subclinical DEB has been shown to be associated with an increase in complications: retinopathy (Rydall 1997), neuropathy (Steel 1987), transient lipid abnormalities (Affenito 1997b), increased hospitalizations for diabetic ketoacidosis (Rodin 1992), and poorer short-term metabolic control (Rodin 1992, Affenito 1997a, Meltzer 2001). Cross-sectional studies have shown associations between elevated A1C and the presence of diagnosable ED (Wing 1986, Affenito 1997a, Herpertz 1998a), subclinical DEB (Wing 1986), and intentional insulin omission (Jones 2006). Associations between DEB and the complications of T2D have not been extensively examined (Herpertz 1998b, Herpertz 2000, Herpertz 2001, Papelbaum 2005). Refusal to initiate insulin treatment by patients with T2D (psychological insulin resistance) (Davis 2006) may be driven, in part, by concerns about weight gain, but this is anecdotal and has not been systematically tested.
An assessment of DEB should be performed when weight gain, weight loss, and/or worsening glycemic control (including severe hypoglycemia and/or ketoacidosis) cannot be explained by disease processes, changes in care, medication or insulin regimen, a monitored weight-loss program, obvious noncompliance, or psychiatric morbidity, especially in young women (Daneman 2002).
Prevalence of Diagnosable ED and Subclinical DEB
Diagnosable ED has low prevalence in the diabetes population (Ackard 2008). There are varying estimates of the prevalence of diagnosable ED and DEB in individuals with T1D compared with healthy referent populations (Crow 1998, Engstrom 1999). Estimates range from 3.8% (Pollack 1995) to 31% in adolescent and young adult women with T1D (Polonsky 1994). Some studies have found similar rates to the general population and some higher, but assessment methods vary (Rodin 1986–1987, Steel 1989, Fairburn 1991, Peveler 1992, Striegel-Moore 1992, Crow 1998, Herpertz 1998a, Bryden 1999, Engstrom 1999, Meltzer 2001).
Subclinical DEB is increasing in all segments of the U.S. population and Westernized cultures, presumably associated with emphasis on the thinness ideal and concern about overweight/obesity. Prevalence rates of subclinical DEB may be underestimated because dieting behavior is common and there is a stigma attached to self-reporting DEB (Neumark-Sztainer 2002b). The prevailing belief is that the diagnosis of diabetes is associated with elevated rates of DEB when intentional reduction in insulin dose or omission is considered purging behavior to control weight, especially in women with T1D and in adolescent girls (Hudson 1983, Bubb 1991, Rodin 1991, Hockey 1993, Biggs 1994, Pollack 1995, Crow 1998, Affenito 2001). However, a recent study using a population-based healthy comparison sample did not show elevated rates of diagnosable ED in the cohort with diabetes (Ackard 2008).
Bulimic behaviors and insulin omission are the most commonly reported DEB in patients with T1D (Goebel-Fabbri 2008, Alice Hsu 2009), whereas caloric restriction/restraint and binging are more commonly reported by women with T2D (Herpertz 2001, Young-Hyman 2010). Rates of DEB in boys with diabetes have been shown to be considerably lower than those found for women (Meltzer 2001, Neumark-Sztainer 2002a) but may be increasing (Svensson 2003). Higher prevalence of T2D in minority populations could potentially be associated with increased rates of DEB but this relationship has not yet been demonstrated. Studies that compare occurrence of DEB in patients with type 1 versus type 2 diabetes show similar rates; however, types of behaviors reported differ. “Drive for thinness” and “body dissatisfaction” are more common in individuals with T2D. Intentional insulin omission (to cause glycosuria) is more common in patients with T1D (Herpertz 1998b, Herpertz 2001).
Among overweight young women attempting weight loss, both with (type 1) and without diabetes, weight status is a strong predictor of DEB (Striegel-Moore 1992, Vamado 1997, Arriaza 2001, Sherwood 2001, Neumark-Sztainer 2002b, Rodin 2002, Decaluwe 2003, de Man Lapidoth 2006, Shisslak 2006). In studies reporting BMI, type 1 cohorts have been significantly heavier than healthy control subjects, with average BMI above the normal range. Elevation in weight, independent of diagnosis, would in itself predict higher rates of DEB (Engstrom 1999, Jones 2006). However, few studies of diabetes cohorts have compared rates of DEB with healthy control groups matched for age, sex, and weight (Engstrom 1999, Colton 2004, Jones 2006). When subjects matched for BMI were used to compare overweight and obese patients having T2D with obese nondiabetic patients seeking weight loss and an obese nonclinical sample (Battaglia 2006), low levels of binge eating disorder were diagnosed overall (<5% in all groups). However, obese patients with diabetes had the lowest scores on the Eating Disorder Examination, but the highest scores on the Restraint scale (Mannucci 2002). Higher scores on the Restraint scale were attributed to treatment behaviors. Although there is robust documentation that behavior generally considered subclinical DEB using DSM criteria such as binge eating, purging (defined as intentional insulin omission), and caloric restriction are commonly reported by patients with diabetes (Fairburn 1991, Bryden 1999, Jones 2006), it is not known how much these reports reflect cognitions based on attempted adherence to the diabetes care regimen.
Weight gain consequent to good blood glucose control could be a driver of weight concerns (Meltzer 2001, Battaglia 2006) and is known to be a side effect of successful treatment (Steel 1990). Although the presumption is that DEB in this population is associated with elevated BMI levels, only one study stratified the diabetes cohort (both type 1 and 2 diabetes, men and women, age range 18–65 years) by weight status. Three percent of under- and normal weight women had a current ED, whereas 6.8% of the overweight and 10.3% of obese women reported DEB (Herpertz 1998b). These rates are similar to samples with equivalent BMIs seeking weight loss (Vamado 1997). A conflict may exist between the need to control weight and achieve good glycemic control (in patients with both T1D and T2D). In particular, young adult and adolescent women have been shown to use insulin omission specifically for weight control (Biggs 1994, Khan 1996). Fear of improved glycemic control “because I will gain weight” and diabetes-specific distress predict intentional insulin omission (Polonsky 1994). Although one goal of medical nutrition therapy (MNT) is to prevent weight gain (Nathan 2005), supervised weight management programs are not routinely available when weight gain occurs secondary to successful treatment with insulin.
Etiology of ED and DEB in the Diabetes Population: Psychiatric Symptoms, Regimen Compliance (or Noncompliance), or Physiologic Dysregulation
Primary risk factors for ED and DEB (in the nondiabetic population) are weight and size concerns, early eating problems and dieting, the presence of other forms of psychopathology, sexual abuse and other adverse life experiences, and low self-worth (Jacobi 2004). Except for weight concerns and depression, the relationships between these risk factors and occurrence of DEB in the diabetic population have received little attention.
Establishing the occurrence of DEB attributable to having diabetes and managing the disease is complicated by the paucity of studies that concurrently assess psychiatric symptoms, psychological adjustment to illness, and sequelae of the diabetes care regimen. Bryden et al. (1999) tracked BMI along with weight and shape concerns in adolescents and young adults with T1D; as both men and women became overweight, DEB increased. However, baseline and ongoing psychological status, independent of weight concerns, was not assessed. In contrast, Pollock et al. (1995) followed new-onset girls and boys with T1D (ages 8–13 years) from diagnosis for up to 14 years. The presence of DEB, compliance with medical regimen, and psychiatric symptoms were assessed, including weight concerns. Low rates of DSM-III diagnosable ED (3.8%) were found; however “youths with eating problems were nine times