| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
EDITORIAL COMMENT |
Department of Psychiatry, Washington University School of Medicine, Department of Veterans Affairs Medical Center, St. Louis, Missouri (Lustman)
Department of Medicine, Washington University School of Medicine, St. Louis, Missouri (Clouse)
The association of depression with diabetes was first noted in the scientific literature more than 300 years ago when Willis made the surprising remark that diabetes was the result of sadness or prolonged sorrow (1). After that time, investigators have studied this comorbidity in patients with either diabetes or depression as the primary disorder and focused on the prevalence of depression in patients with either type of diabetes, the relationship of depression with measures of diabetes severity (insulin resistance (IR), hyperglycemia, and complications), the risk of developing Type 2 diabetes in depressed patients, and the effect of depression on diabetes-related mortality.
Data from two of these principal lines of investigation have led scientists to conclude that a) the prevalence of depression is increased significantly in persons with diabetes (Type 1 or Type 2) compared with those without diabetes (2,3) and b) depression appears to increase significantly the likelihood of developing Type 2 diabetes (4). The latter effect is independent of other conventional risk measures and presumably results from direct effects of depression on glucose dysregulation preceding the ultimate diagnosis of diabetes (5,6). Both conclusions are supported by meta-analyses (3,4). One such analysis (4), an aggregate of data from nine longitudinal studies having follow-up periods of 3 to 16 years, concluded that depressed adults have a 37% increased risk of developing Type 2 diabetes. Outcomes from these lines of investigation are very important to patients with or at risk for diabetes in that they may position depression as a potentially modifiable factor influencing the onset and course of the medical illness.
In this issue of Psychosomatic Medicine, Knol et al. further contribute to this story with a cross-sectional study examining the rate of increased depression symptoms in patients self-reporting the diagnosis of diabetes (7). The authors also examine this rate in subsets with increased fasting blood sugar levels that did not meet the criteria for diabetes and that did, but subjects in the latter subset were unaware that they had the disorder. Only patients who carried the diagnosis of diabetes had greater risk of increased depression symptoms (in the unadjusted comparisons), but the significance of this observation was lost when all potential confounders, including other medical comorbidities, were included. Because the confidence interval crossed one with the final adjustment, the authors were led to conclude that the association of depression symptoms with diagnosed diabetes in the univariate analysis was related to the burden of medical illnesses in these patients and not to diabetes specifically. They were unable to demonstrate that acute depression symptoms were linked directly to either diabetes (diagnosed or undiagnosed) or prediabetes—findings that would speak against both conclusions from existing lines of investigation.
Why are these outcomes at odds with prior work? The present study assessed depression with a self-report measure of current symptom severity, not an interview-determined Diagnostic and Statistical Manual of Mental Disorders diagnosis, and with antidepressant use as a surrogate. These methods have limited predictive value for major depression and interfere with the correct segregation of subjects (8). In a meta-analysis of 24 studies that showed the association of depression with hyperglycemia (9), effect sizes were larger when standardized interviews and diagnostic criteria rather than self-report questionnaires were used to assess depression (0.28 versus 0.15). But even those previous studies that relied on depression symptoms alone demonstrated a two-fold increase in the depression rate in subjects with diabetes compared with the control group (3). Is it possible that the bulk of this association is not attributable to diabetes per se but is created spuriously from the burden of additional medical illnesses often seen in diabetic patients?
Speaking to this, Gavard et al. (2) found in a critical review that few studies considered comorbid medical illness when evaluating the prevalence of depression in diabetes, and, when they did, the rates of other medical conditions were higher in the diabetic subjects. Multivariable adjustment for comorbid illness had not been performed, but three studies compared the rates of depression in diabetic patients with those in medically ill control groups (10–12)—another way of approaching this question. Weyerer et al. (10) found an increased rate of structured interview-ascertained major depression in a Type 2 diabetes sample, although it did not reach statistical significance (27.3% versus 20.3%). Two other studies (11,12) found significantly greater increases in depression symptom scale scores in diabetic subjects than in medically ill controls. Subsequently, using a large sample from a health maintenance organization, Nichols and Brown (13) found a higher rate of diagnosed depression in a sample with Type 2 diabetes than in a control group after adjusting for the presence of cardiovascular disease.
Although it is very likely that experiencing other medical illnesses contributes to depression and partially explains previous conclusions (3,14), this factor appears insufficient to negate the association when specifically examined. Possibly the use of current depression symptoms, reliance on self-reported diagnoses, and low participation rate (of those invited, 50% declined) interfered with the detection of a conclusive, specific diabetes-depression association in the current study by Knol et al.
This issue would be less disconcerting if significant direct relationships of depression symptoms with forms of hyperglycemia that precede the diagnosis of Type 2 diabetes (prediabetes, undiagnosed diabetes) had been demonstrated. Receiving the diagnosis of diabetes further compounds the medical burden, but are diagnoses and illness burdens all that factor into previously described observations of the doubled rate of depression in diabetes samples? The predictive value of depression for subsequent development of Type 2 diabetes over longitudinal observation, as supported by meta-analysis, would speak against this (4). Many of these studies adjust for the severity of comorbid medical illness. One study (6) measured IR, a precursor to Type 2 diabetes, using homeostasis model assessment (HOMA-IR). HOMA-IR was significantly higher in the subset with increased depression symptoms after controlling for age, race, site, education, and use of medications for depression or nervous conditions.
IR resulting from depression is a leading suspected mechanism linking depression to Type 2 diabetes in longitudinal studies (15). Depression-associated insulin resistance (DAIR) could mediate exaggerated hyperglycemia in depressed patients with either Type 1 or Type 2 diabetes (5,15), may explain the increased risk of coronary heart disease development referable to depression in diabetic women (5,16), and also has been observed in nondiabetic persons (18–20) with unipolar or bipolar depression (20,27). The process likely involves multiple, interactive mechanisms, including depression-related factors known to increase IR, such as physical inactivity (28–30), cortisol and inflammatory cytokine elevations (5,31), and obesity (32–34), particularly obesity in women (35)—although weight-independent mechanisms are also operational (36).
Reduction in IR or hyperglycemia with depression treatment also supports the DAIR hypothesis. Okamura et al. found that IR measured from frequently sampled intravenous glucose tolerance testing in nondiabetic depressed subjects improved with imipramine therapy (18). Treatment of major depression in diabetic patients with pharmacotherapy or cognitive behavior therapy produced significant improvement in glycemic control in most (36–40) albeit not all (41) randomized, controlled, acute- and maintenance-phase trials thus far, the effects being unexplained by adherence to diabetes management or weight change, when measured. Up to this time, however, changes in IR have not been measured specifically in depression treatment studies of diabetic patients.
Our final question: Does the present study by Knol et al. burst the bubble on the DAIR story? Limitations related to sample selection, statistical power (only 11 subjects had depression in the undiagnosed diabetes group), imprecision in the identification of depression cases, uncertainties inherent to assuming that a fast preceded single glucose testing, and limited evaluation of the spectrum of glucose regulation abnormalities (e.g., impaired glucose tolerance) reduce the strength of any conclusion. As disappointing as these data seem in supporting the role of depression as a risk factor for Type 2 diabetes, they remind us of the importance of the burden of medical illness, including diabetes, in precipitating depression. The latter may be more easily demonstrated while interfering with the detection of an inverse relationship, and rigorous attention to methodological details, particularly depression histories (e.g., family history, age of onset, time spent in depression), will likely be necessary to conclusively solve the puzzle. For now, we adhere to the findings from longitudinal and treatment studies and look forward to future controlled investigations that measure IR directly. Detecting modifiable risk factors for diabetes and its complications is a process with very high stakes for the community with diabetes and for society. Studies must be interpreted carefully and within an entire investigative framework to ensure that opportunities for discovery are not put to bed prematurely.
DOI:10.1097/PSY.0b013e318060cc2d
 |
REFERENCES |
|---|
 TOP REFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. A. Zauszniewski and G. C. Graham Comparison of Short Scales to Measure Depressive Symptoms in Elders With Diabetes West J Nurs Res, March 1, 2009; 31(2): 219 - 234. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
