Lustman PJ, Anderson RJ, Freedland KE, de Groot MK, Carney RM, Clouse RE:Depression and poor glycemic control: a meta-analytic review of the literature. Diabetes Care 23:934–942, 2000

Evidence that depression is associated with hyperglycemia and diabetes spans case reports and series, cross-sectional studies, and randomized clinical trials. Several reports describe peripheral insulin resistance and hyperinsulinemia in individuals with depression.16–18 Chiba et al.19 reported that depressed nondiabetic patients had decreased insulin sensitivity compared to nondepressed nondiabetic controls. Okamura et al.20 studied 20 nondiabetic depressed patients with oral and frequently sampled intravenous glucose tolerance tests before and after treatment with tricyclic antidepressants. They observed that non-diabetic depressed patients had decreased insulin sensitivity compared to nondepressed nondiabetic control subjects. After treatment, the patients with depression showed significant increases in insulin sensitivity. Lustman et al. performed this metaanalytic review of the literature in order to help clarify whether depression is associated with poor glycemic control.

Studies published within the 25-year period before the meta-analysis were screened for inclusion. Included studies were limited to those that used adult participants and collected hemoglobin A1c (A1C) and depression information at the time of study assessment. Studies were excluded from the meta-analysis if they had fewer than 25 participants, were not available in English, or collected only a history of depression. Using established meta-analytic techniques, an effect size (ES) was calculated for each study. Because of the heterogeneity of design methods, studies were divided into two subsets: cross-sectional (n = 26) and randomized clinical trials(n = 5).

Of the 26 cross-sectional studies, 10 studied patients with type 1 diabetes, 6 studied type 2 diabetes, and 10 studied a mixed sample. Twenty-four studies covering 2,817 subjects provided enough information to calculate the ES. The combined effect showed that depression was significantly associated with hyperglycemia (P < 0.0001). The overall ES was 0.17, and it was greater in studies that obtained interview-based diagnoses rather than relying on patient self-report measures (ES 0.28 vs. 0.15).

The five randomized clinical trials observed the relationship between depression and glycemic control over time during experimental control of one of these two variables. Three of the studies examined the effect of cognitive behavior therapy, nortriptyline, or fluoxetine treatment of depression on glycemic control.21–23 In two of the three studies, glycemic control improved with treatment,22,23 and in two of the three studies reduction in depression severity was associated with improved glycemic control.21,22 The other two studies, observed improvements in depression related to treatment of diabetes.24,25

Although the authors note that the overall ES observed in these studies is small, they point out that treatment of depression in a diabetic population could potentially bring the proportion of patients with diabetes who are in good control from 41 to 58%. Unfortunately, little is known about the sequence of events or the host and environmental factors that lead to the association between depression and diabetes. The question that remains in light of these findings, however, is the extent to which the biology of depression per se might increase the risk for impairments in glucose metabolism versus the effects of depression-related host factors (e.g., behavioral changes) or depression-related environmental factors that could contribute to the association with impairments in glucose metabolism.

Hormonal abnormalities associated with the hypothalamic-pituitary-adrenal axis have been associated with depression, including hypercortisolism.26–29 It has been hypothesized that depression may be associated with insulin resistance secondary to hypercortisolism, based on the effect of elevated plasma glucocorticoid levels to decrease insulin sensitivity.18,30 However, these studies of glucose regulation in depression have generally failed to characterize subjects' level of adiposity, activity level, caloric intake, family history of diabetes, or the longitudinal course of changes in insulin secretion or action.

Although positron emission tomography studies of individuals with depression have described a pattern of impaired central nervous system glucose metabolism,31,32 it is not clear how this is related to peripheral changes in glucose metabolism. In vitro evidence indicates that circulating insulin levels can regulate glycogen storage within brain glia.33In addition, in vivo human studies support a relationship between chronic peripheral changes in glucose metabolism and central nervous system cognitive responses to an acute glucose challenge.34 To a large degree,however, the relationship between peripheral changes in glucose metabolism and central changes in brain energy metabolism remains unexplored.

Little is known about the behavioral and environmental factors associated with depression that may relate to risk of hyperglycemia and diabetes, such as socioeconomic status, sedentary lifestyle, diet, smoking, and others. This kind of information is important to evaluating the relative contributions of depression per se, related behavioral changes, and environmental factors to the development or worsening of hyperglycemia in certain individuals. The observation that other major psychiatric conditions, such as schizophrenia,are similarly associated with an increased risk of diabetes and related cardiovascular mortality compared to the general population35suggests that common factors, such as changes in nutrition, activity level,socioeconomic status, and access to medical care, may be important. Identification of these factors will be critical to directing public health efforts at risk reduction.

Accurate targeting of primary and secondary prevention efforts may be even more important in those patients with depression or other psychiatric conditions where pharmacotherapy (e.g., certain antipsychotic medications) can make an adverse contribution to disturbances in glucose and lipid metabolism. Treatment-induced increases in adiposity along with related, and in some cases unrelated, changes in glucose and lipid metabolism can contribute to overall risk. This increasingly recognized iatrogenic problem increases the importance of identifying all the other factors that can increase risk for patients,particularly when changing medications is not an option. Given the combined public health significance of major mental illness and diabetes, and the association between the two, further research is necessary to identify modifiable risk factors, characterize their contributions to disease, and design optimal interventions that maximize health outcomes.

Dan Haupt, MD, is an instructor in psychiatry, and John Newcomer, MD,is an associate professor of psychiatry at Washington University School of Medicine in St. Louis, Mo.

Note of disclosure: Dr. Haupt has received research support from Abbott Laboratories and honoraria from Abbott, AstraZeneca Pharmaceuticals,Bristol-Myers Squibb (BMS), Janssen Pharmaceuticals, Pfizer Inc., and Sanofi-Synthelabo. He has also served as a consultant for Abbott, AstraZeneca,BMS, and Janssen. Dr. Newcomer has received research support from Janssen, Eli Lilly and Company, Pfizer Inc., AstraZeneca, and BMS. He has also served as a consultant for Janssen, Eli Lilly, Pfizer, AstraZeneca, BMS, and GlaxoSmithKline, and he has received honoraria from Janssen, Pfizer,AstraZeneca, BMS, and Sanofi-Synthelabo. These companies manufacture products for the treatment of depression and other affective disorders.