Deconstructing Diabetes and Depression: Clinical Context, Treatment Strategies, and New Directions

Diabetes mellitus is characterized by chronic hyperglycemia that in the absence of treatment leads to microvascular (e.g., neuropathy, nephropathy, retinopathy) and macrovascular (e.g., myocardial infarction, stroke, amputation) complications (6). Type 1 diabetes (T1DM) is an autoimmune disorder that generally presents before adulthood and leads to a total insulin deficiency. T2DM is a consequence of insulin resistance and a relative insulin deficiency; it is more prevalent and presents later in life (6). People with impaired fasting glucose or glucose tolerance—sometimes referred to as prediabetes—are at high risk of progressing to T2DM (6).

Unlike diabetes, which is a laboratory-based diagnosis, depression is a heterogeneous concept defined by the presence and duration of specific symptoms. Major depressive disorder is a chronic mood disorder characterized by low mood, anhedonia, neurovegetative symptoms (e.g., disturbance of sleep, appetite, energy, and concentration), as well as negative thought processes (e.g., excess guilt, worthlessness, and suicidal ideation) (3). A major depressive episode, occurring when a threshold of these symptoms is reached for a 2-week period, is the hallmark of major depressive disorder. Major depressive events are not unique to major depressive disorder (e.g., major depressive events are also typical of bipolar disorder), and subthreshold depressive symptoms frequently occur among those without major depressive disorder (7).

For research purposes, validated self-report symptom scales (e.g., Beck Depression Inventory, Patient Health Questionnaire–9 [PHQ-9]) are used to screen for depression. The gold standard for diagnosing depression, however, is the Structured Clinical Interview for DSM-IV Axis I Disorders, which is more specific but also more resource-intensive (7). This review refers to clinician-diagnosed depression as major depressive disorder and self-report measures as depressive symptoms whenever the distinction is required.

Fifteen years ago, Anderson et al. determined that the presence of diabetes doubles the odds of depression (8). Subsequent studies affirmed the initial findings (8–11). In a 2006 meta-analysis of cross-sectional studies, depression had a prevalence of 17.6% among those with T2DM and 9.8% among those without T2DM (odds ratio [OR]=1.6, 95% confidence interval [CI]=1.2–2.0) (9). However, diabetic patients differed from the nondiabetic patients on other variables known to be associated with an increased risk of depression (e.g., comorbid chronic illnesses) (9). In addition, the prevalence of comorbid depression among patients with diabetes depends on the assessment method, ranging from 11% found with the standardized interviews to 31% with self-report questionnaires (8, 10). Comorbid depression and diabetes are particularly common among women, with female preponderance noted in the 2001 findings (28% for women vs. 18% for men) and 2006 meta-analysis (23.8% vs. 12.8%) (8–10).

For patients with T1DM, meta-analysis of cross-sectional studies showed a 12% prevalence of depression, in comparison with 3.2% among control patients without T1DM, but heterogeneity limited conclusions (12, 13). Subsequent cross-sectional studies support an increased prevalence of depression among adults with T1DM, but the evidence is inconclusive for the under-25-year-old population (11–15).

In addition to female sex, established risk factors for depression are also risk factors for depression in the diabetic population (e.g., older age, poor social support, and low socioeconomic status) (9, 11). Additional risk factors for depression specific to the diabetic population include occurrence of late or acute complications, persistently poor glycemic control, insulin therapy initiation in T2DM, and the perception of deteriorating health (9, 10). A troubling finding is an increasing prevalence of depression in T2DM among hospitalized patients, going from 3.5% in 2001 to 5.8% in 2011 (16).

Recently, scales such as the Diabetes Distress Scale and Problem Areas in Diabetes Scale have measured diabetes distress, which reflects the emotional response to the burden of diabetes and its complications, as well as the prolonged psychological distress during adjustment to the diagnosis (8). Diabetes distress presents in its severe form to 10%−30% of people with diabetes (8). Measures of diabetes distress correlate strongly with self-reports of depressive symptoms (coefficient: 0.22–0.63) but less so with standardized diagnostic interviews (coefficient: 0.04–0.15) (8, 15). In the T1DM population, use of the PHQ-9 has led to a high false-positive rate (52%−71%) compared with use of the Structured Clinical Interview for DSM-IV Axis I Disorders, suggesting that some of what was considered depression among adults with diabetes might represent diabetes distress (15). Although depression and diabetes distress are partially overlapping constructs, they are not interchangeable.

A 2015 meta-analysis examined the prevalence of T2DM among patients with diagnosed major depressive disorder (17). Patients with major depressive disorder were more likely to have T2DM (OR=1.49, CI=1.29–1.72), even when analyses controlled for age and sex (OR=1.36, CI=1.28–1.44) (17). In the largest of the included studies (143,943 primary care patients with major depressive disorder), T2DM was present in 9.3% of the major depressive disorder population, in comparison with 4.3% of those without depression (18).

Building on the cross-sectional associations, longitudinal studies inform on the temporal relationship between diabetes and depression in order to determine whether one might be a risk factor for the development of the other, or vice versa. In 2006, a meta-analysis of nine longitudinal studies found that patients with either a diagnosis of depression or depressive symptoms were at 37% increased risk of developing T2DM, with a pooled relative risk of 1.26 (CI=1.13–1.39) (19). This finding was echoed in 2008, when a meta-analysis of 13 longitudinal studies found that baseline major depressive disorder diagnosis or presence of depressive symptoms was associated with a 60% increased risk of T2DM (pooled relative risk of 1.60, CI=1.37–1.88) (20). A 2014 meta-analysis reaffirmed these findings, noting a pooled relative risk of 1.32 (CI=1.18–1.47) (21). For patients with depression the risk of developing T1DM remains unclear, owing to the natural history of the two illnesses (with the onset of T1DM preceding the onset of major depressive disorder in most cases).

The increased risk of T2DM for patients with depression may relate to behavioral factors (22–25) (see Figure 1). Depression is associated with higher body mass index, poor diet, less physical activity, and smoking, all of which are risk factors for T2DM (22). This hypothesis posits that the symptoms of depression (e.g., anhedonia, sleep and appetite dysregulation, fatigue) lead to an unhealthy lifestyle and poor self-care, which in turn increases the risk of obesity and T2DM (23). In support of this, a 2008 study found higher rates of new-onset T2DM among those with elevated depressive symptoms, but after analyses controlled for lifestyle factors such as smoking, daily caloric intake, physical activity, and alcohol use, the trend for increased T2DM risk was no longer significant (26).

A second hypothesis explaining the increased T2DM vulnerability of patients with depression involves the biochemical changes associated with depression (see Figure 1). Depression is associated with dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis and sympathetic nervous system, particularly in the severe or melancholic subtypes of depression (27, 28). Excess production of glucocorticoids by the adrenal glands can contribute to metabolic syndrome, insulin resistance, and T2DM (27, 28); therefore, overactivation of the HPA axis might explain the T2DM-depression relationship. In addition, evidence from meta-analysis demonstrates that depression is associated with higher circulating levels of proinflammatory cytokines such as tumor necrosis factor and interleukin-6 (29). Raised concentrations of inflammatory cytokines, specifically interleukin-6, have been associated with increased risk of T2DM (30). Lastly, depression may have a direct impact on insulin resistance (24). A recent meta-analysis found a significant association between depression and insulin resistance (effect size=0.19, CI=0.11–0.27), with a larger effect size when standardized interviews were used for diagnosis (31). Furthermore, some drugs that reduce insulin resistance, such as pioglitazone, improve depressive symptoms (24). However, the relationship between depressive symptoms and other glycemic traits (e.g., fasting glucose, glucose tolerance) has been inconclusive (24).

Antidepressant medication is a potential confounding factor in the relationship between T2DM and depression, because antidepressant use has been associated with increased risk of physician-diagnosed T2DM (32, 33). However, antidepressant use is not associated with changes in fasting glucose or glucose tolerance, nor is it associated with undiagnosed diabetes (32, 33). The finding that antidepressants are associated with T2DM diagnosis but not increased glucose concentrations suggests a noncausal relationship, with the detection/ascertainment bias a proposed explanation (e.g., increased contact with medical care leads to improved detection, or diabetes may be detected when ruling out endocrine causes of depression) (33).

Multiple meta-analyses of longitudinal studies have determined that people with T2DM have an elevated risk of depression. According to a 2008 meta-analysis of seven studies, the pooled relative risk for incident depression in the population with baseline diabetes was 1.15 (CI=1.02–1.30), corresponding to a 15% increased risk (20). In 2010, analysis of 11 studies found a pooled relative risk of 1.24 (CI=1.09–1.40) (34); a 2012 meta-analysis of 16 studies reported an adjusted relative risk of 1.25 (CI=1.10–1.44) (35). Of note, the increased risk was significant in studies measuring depression with diagnostic interview or antidepressant usage but not in studies using questionnaires (35). To date, no meta-analysis has reported a relative risk of depression in T1DM.

One explanation is that the psychological burden of a diagnosis predisposes patients to depression. The psychological burden is a multifaceted construct, encompassing disease-related cognitions (e.g., perceived disability, awareness of illness) and the interaction of increased self-care demands with social support and coping style (36). A 2011 meta-analysis found that individuals with undiagnosed diabetes had a significantly lower risk of depression than did those with previously diagnosed diabetes (OR=0.57, CI=0.45–0.74) (37). In addition, the risk of depression did not differ between individuals with undiagnosed diabetes and those with either impaired glucose metabolism or normal glucose metabolism (37). This suggests that the diagnosis itself, rather than the biochemical changes leading up to diagnosis, underlies the increased depression risk. Two studies have noted an increase in antidepressant prescription in the first year after T2DM diagnosis, consistent with an acute worsening of mental health following T2DM diagnosis (38).

A second possibility is that the structural and neurochemical changes associated with diabetes induce depressive symptoms (22, 24, 36). The neuroinflammatory changes induced by diabetes and the direct impact of altered insulin metabolism on the brain are biologically plausible mechanisms by which those with diabetes could be predisposed to depression (39–41). Diabetes-related biochemical changes induce cognitive disturbances and can adversely affect emotional processing and reward circuitry (41–43).

In advanced diabetes, the complications of the illness (and resulting self-care needs or disability) might cause an increased risk of depressive symptoms (4, 44). Some of the complications of diabetes—such as cardiac disease, cerebrovascular disease, kidney disease, or poor vision—are themselves independent risk factors for depression (44). Thus, the increased risk of major depressive disorder in diabetes is mainly attributed to psychosocial factors, although there is increasing evidence for a biochemical mechanism (see Figure 1).

Depression in diabetes leads to suboptimal self-care and poorer glycemic control, leading to an increased risk of complications and mortality (45). There is a modest yet significant association between depressive symptoms and hyperglycemia among diabetic patients, supported by meta-analysis of cross-sectional studies and subsequent longitudinal studies (11, 45, 46). However, recent prospective studies suggest that diabetes distress, rather than depressive symptoms, might be a better predictor of suboptimal glycemic control (15, 45, 47). Among patients with diabetes, those with comorbid depressive symptoms are less active, more likely to smoke, less likely to eat healthfully, and less likely to adhere to treatment (including glucose monitoring and antidiabetes medication) (48). A 2008 meta-analysis of 47 studies found a significant association between depression and nonadherence to diabetes treatment; the effect size was highest for missed appointments and composite measures of self-care (49). In this case, depressive symptoms had a stronger negative relationship with self-care than did diabetes distress (50). In a recent longitudinal study, persisting depressive symptoms or worsening depressive symptoms at 5-year follow-up were associated with worse adherence to diet and exercise regimens, whereas patients with baseline depression who had clinical improvement in their depression were no different from the nondepression group with regard to diet and exercise adherence (51). Higher levels of depressive symptoms or diabetes distress are associated with negative beliefs about insulin among insulin-naïve T2DM patients, suggesting that depression might delay the initiation of insulin in T2DM (52).

Among patients with diabetes, depression has been associated with a higher prevalence of both microvascular and macrovascular complications (53). Depression among patients with diabetes has been associated with a higher incidence of blindness, diabetic foot ulcers, coronary artery disease, end-stage renal disease, and amputations (45).

Evidence also suggests that diabetes among patients with depression leads to poorer depression outcomes (54). When diabetes is comorbid, the course of depression appears to be of longer duration, with more recurrences and greater severity (55). Even with successful treatment, relapse may occur with as much as 80% of patients with diabetes with major depressive disorder (55). Poor metabolic control can exacerbate depression or diminish response to antidepressant medication (56). In addition, those with depression notice the symptoms of diabetes more and perceive a greater burden of illness (57).

Patients with depression and diabetes are at greater risk of dementia (hazard ratio [HR]=4.84, CI=4.21–5.55), and the excess risk to those with both conditions is higher than the additive risks of the individual diagnoses (58). Cognitive dysfunction results from both depression and diabetes, and mild cognitive impairment is more likely to proceed to Alzheimer’s dementia if either of the two diagnoses is present (59). There is a complex neurobiological relationship between psychiatric disorders (e.g., major depressive disorder) and metabolic changes (e.g., T2DM), which leads to synergistically adverse changes in cognitive processes (e.g., alterations in brain connectivity and impairment of brain plasticity) (39, 41–43).

Patients with both diabetes and depression have an 82% increased risk of myocardial infarction (HR=1.82, CI=1.69–1.97) in comparison with those with neither condition (60). Notably, a recent meta-analysis of 16 prospective studies demonstrated that baseline depression among patients with diabetes was associated with nearly 50% increased risk of all-cause mortality (HR=1.46, CI=1.29–1.66), mainly attributed to cardiovascular mortality (HR=1.39, CI=1.11–1.73) (61).

Depression is underdetected among diabetic patients (54, 63, 64). In accordance with this, multiple clinical guidelines advocate for increased screening for depression among patients with diabetes (65–67). Routine screening for anxiety, eating disorders, and cognitive impairment is also recommended (67). Similarly, screening for T2DM (e.g., measuring fasting glucose, measuring hemoglobin A1c) and identification of those at risk (e.g., glucose tolerance testing, monitoring for metabolic syndrome) should be fundamental aspects of treatment for depression, as indicated by recent guidelines (68, 69). Screening for depression typically utilizes depression self-report questionnaires with validated cut-off scores. The Beck Depression Inventory, the Center for Epidemiologic Studies Depression Scale, the Hospital Anxiety and Depression Scale, and the PHQ-9 have been validated in diabetic populations (70–72). Nonetheless, the PHQ-9 has the strongest evidence of validity (71).

Unfortunately, these scales are limited in specificity and have a tendency to overestimate rates of depression, particularly given the potential overlap between diabetes-related symptoms and depression symptoms (70, 71, 73). For example, cognitive dysfunction may be a primary complication of diabetes or a symptom of depression, and other symptoms such as fatigue or appetite changes can present in both disorders (74). Furthermore, there is considerable overlap between depression and diabetes distress, and recent efforts to differentiate the two have led to the development of scales such as the Diabetes Distress Scale and Problem Areas in Diabetes Scale (7). Optimization of screening protocols is underway to improve specificity while maintaining adequate sensitivity (71). One solution is to increase the scales’ cut-off values. For example, the best cut-off score for the PHQ-9 in a diabetic population is 12 (rather than 10 in the general population) (75).

Regardless of the screening tool used, positive screens require a subsequent diagnostic interview with the intention of confirming or refuting the diagnosis. Primary care providers frequently diagnose and treat major depressive events within their scope of practice, so referral to a psychiatrist is often unnecessary. However, it is essential to have adequate follow-up and well-defined treatment pathways, because the evidence has shown that screening for depression on its own does not improve outcomes (76, 77). Screening is only beneficial when adequate follow-up and effective treatments are available (77). Overall, care for those with diabetes and depression must be integrative and multidisciplinary; collaborative care leads to better treatment adherence, greater improvement of depression and glycemic control, and better quality of life (78, 79).

After confirming diagnosis, clinicians should use evidence-based treatment targeting both depressive symptom severity and glycemic control in the depressed diabetic population (62, 80). Many clinical trials have evaluated the effect of interventions on both sets of outcomes (psychological and medical) (64). Categorically, the interventions can be grouped into psychological interventions (e.g., problem-solving techniques, counseling, cognitive-behavioral therapy [CBT]), psychopharmacological treatments (e.g., antidepressants, anti-inflammatories, insulin sensitizers), and lifestyle modification (e.g., diet, exercise) (64).