Eveningness Is Associated With Greater Depressive Symptoms in Type 2 Diabetes Patients: A Study in Two Different Ethnic Cohorts

ABSTRACT

Background: Eveningness is associated with greater depressive symptoms in the general population. Depression and type 2 diabetes (T2D) commonly coexist. We aimed to explore the association between morningness–eveningness and depressive symptoms in T2D patients in the United States and in Thailand. Participants: T2D patients (n = 182) from an endocrinology clinic in Chicago, Illinois, and six hospitals in Thailand (n = 251) were enrolled. Methods: Diabetes history was collected. Depressive symptoms were assessed by the Center for Epidemiologic Studies Depression scale (CES-D). The Chicago cohort completed the Morningness-Eveningness Questionnaire (MEQ) and the Thai cohort completed the Composite Scale of Morningness (CSM). Sleep quality was assessed using the Pittsburg Sleep Quality Index (PSQI). Results: The mean (SD) CES-D score was 13.7 (9.1) in Chicago and 11.9 (6.4) in Thailand. In Chicago participants, after adjusting for age, sex, ethnicity, hemoglobin A1c, insulin use, and PSQI score, greater eveningness (lower MEQ scores) was associated with higher CESD scores (B = –0.117, p = 0.048). In Thai participants, after adjusting for age, sex, and PSQI score, eveningness (lower CSM score) was associated with higher CES-D score (B = –0.147, p = 0.016). In both cohorts, however, eveningness was not independently associated with the likelihood of being in the at-risk range for clinical depression (CES-D ≥ 16). Conclusions: Eveningness is independently associated with greater depressive symptoms in T2D in two different ethnic cohorts. The results support the association between individual differences in circadian rhythms and psychological functioning in T2D.

Diabetes and depression are both common chronic conditions that often coexist. In the United States, diabetes prevalence was estimated to be at 9.3% or 29.1 million people in 2012 (National Center for Chronic Disease Prevention and Health Promotion, 2014), 90–95% of whom had type 2 diabetes, while the 2009–2012 National Health and Nutrition Examination Survey revealed that 7.6% of Americans aged 12 and over had depression (Pratt & Brody, 2014). Beyond being common, there is evidence that diabetes and depression share genetic, biological, and psychological factors and have a bidirectional relationship (Ji, Zhuang, & Shen, 2016; Moulton, Pickup, & Ismail, 2015; Renn, Feliciano, & Segal, 2011). Meta-analyses demonstrate that depressed adults have a 37% increased risk of developing type 2 diabetes compared to nondepressed adults (Knol et al., 2006) and individuals with type 2 diabetes are 24% more likely to develop depression than those without diabetes (Nouwen et al., 2010). It is estimated that 25% of diabetes patients have depression (Anderson, Freedland, Clouse, & Lustman, 2001). Female sex, having diabetes complications, and having other comorbidities are associated with higher risk of depression (Clouse et al., 2003; Tada et al., 2014). Diabetes patients with comorbid depression are more likely to experience adverse outcomes, including a 20% higher risk of incident chronic kidney disease and a 25% increase in all-cause mortality (Novak et al., 2016) .

Morningness–eveningness reflects individual differences in the daily timing of sleep and other activities (e.g., eating, exercise). Eveningness has been associated with depressive symptoms in several large population-based studies (Antypa, Vogelzangs, Meesters, Schoevers, & Penninx, 2016; Kitamura et al., 2010; Konttinen et al., 2014; Merikanto et al., 2013). For example, in a study in the general population in Finland (6,071 participants), those who were evening types were 2.7–4.1 times more likely to have various symptoms of depression (Merikanto et al., 2013). Moreover, in those with established depression, evening types have more severe depressive symptoms than morning types (Muller, Olschinski, Kundermann, & Cabanel, 2016), and being a morning type was associated with having fewer depressive symptoms after an onset of major depression (Selvi et al., 2010). Despite these data, there have been no studies to date exploring whether morningness–eveningness is associated with depressive symptoms in patients with type 2 diabetes.

Environmental and sleep-related factors such as geographic location and sleep quality may play a role in the relationship between eveningness and depression. It is known that those residing in more tropical countries have a greater morning preference than those residing in temperate zones (Randler, Prokop, Sahu, & Haldar, 2014). While the prevalence of depression in patients with diabetes in Thailand appears to be similar to that reported in the United States (Anderson et al., 2001; Thongsai, Watanabenjasopa, & Youjaiyen, 2013), it is unclear whether the association between morningness–eveningness and depressive symptoms differs by location, given the known differences in morningness–eveningness between these areas. Moreover, poor sleep quality is commonly found in diabetes patients and has been shown to be related to depressive symptoms (Sun et al., 2016). Evening types typically have poorer sleep quality and this may also be a contributing factor in their depression (Kitamura et al., 2010).

The purpose of this study was to investigate the relationship between morningness–eveningness and depressive symptoms in patients with type 2 diabetes, adjusting for diabetes-related factors and sleep quality in two different ethnic cohorts: American (location: latitude 42N) and Thai (latitude 13N). We hypothesized that eveningness would be independently associated with greater depressive symptoms, regardless of geographic location.

Method

Participants

The study consisted of two different cohorts of type 2 diabetes patients. The first cohort was recruited from February to May 2012 and consisted of 182 non–shift-working (defined as unemployed or performed day work between 6:00 a.m. and 7:00 p.m.) type 2 diabetes patients being treated in an endocrinology clinic at Rush University Medical Center, Chicago, Illinois, and was a part of the cohort previously described (Reutrakul et al., 2013). The second cohort consisted of 251 type 2 diabetes patients recruited from six hospitals in Thailand (which were less than 200 km apart) from January 2014 to June 2016, and a part of this cohort was previously reported (Reutrakul et al., 2015). This cohort had 191 non–shift-working and 60 evening-shift and night-shift working participants (defined as working shifts that started between 3:00 p.m. and midnight, and ended between 3:00 and 8:00 a.m., for at least 3 months). Eligible participants were those with a clinical diagnosis of type 2 diabetes. Exclusion criteria included pregnancy or significant neurological or physical impairments that required patients to depend on others for daily activity. The protocol was approved by the institutional review boards at Rush University Medical Center, and the Ethical Committee of the Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand.

The data collection procedure was the same in all locations. After obtaining written informed consent, weight was measured. Age, height, current medications, and most recent hemoglobin A1c (HbA1c) values (within the prior 3 months) were extracted from patient medical records. Hemoglobin A1c reflects an overall glucose control in the preceding 3 months and is considered a gold-standard measurement of glycemic control, with a level of ≤ 7% being an optimum level in most patients. Participants were interviewed and medical records were reviewed to obtain diabetes history, medication use, and diabetes complications (retinopathy, neuropathy, nephropathy, coronary artery disease, and peripheral vascular disease). Complications were categorized as none or ≥ 1. Additional evaluations are outlined below.

Depressive symptoms assessment

Depressive symptoms were assessed using The Center for Epidemiological Studies Depression (CES-D) questionnaire (Radloff, 1977). This 20-item scale measures symptoms of depression over the past month. Individual scores, from none of the time (0) to all the time (3), are summed to calculate a total score, which ranges from 0 to 60, with higher scores reflecting greater depressive symptoms. A score of 16 or higher is suggestive of clinical depression (Radloff, 1977). The questionnaire was created in English and has been validated in Thai (Trangkasombat, Larpboonsarp, & Havanond, 1997).

Morningness–Eveningness assessment

The Morningness–Eveningness questionnaire (MEQ; Horne & Ostberg, 1976) was used to assess circadian timing preference in participants in Chicago. This questionnaire consists of 19 multiple-choice questions in which participants rate a series of hypothetical situations to determine their temporal preferences for specific activities (e.g., sleep, eating, exercise, and so forth). The total score ranges from 16 to 86, with higher scores indicating greater morningness.

The Composite Scale of Morningness (CSM) assessed circadian phase preference of participants in Thailand. This measure was chosen over the MEQ for this cohort because the Thai version of CSM has been validated (Pornpitakpan, 1998; Smith, Reilly, & Midkiff, 1989). It is composed of 13 items that query the participant’s preferred wake-up and bedtime, preferred time for physical and mental activity, and subjective alertness. The total score ranges from 13 to 55, with higher score indicating greater morningness.

Sleep assessment

The Pittsburgh Sleep Quality Index (PSQI) has 19 self-rated questions and assesses sleep quality during the previous month (Buysse, Reynolds, Monk, Berman, & Kupfer, 1989). The items form seven component scores (sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbance, use of sleeping medication, and daytime dysfunction), each of which has a range of 0–3 points. A score of 0 indicates no difficulty, whereas a score of 3 indicates severe difficulty. These component scores are summed to yield one global score with a range of 0 to 21, with 0 indicating no difficulty and 21 indicating severe difficulty in all areas. A score of > 5 reflects poor sleep quality. The PSQI has been validated in Thai (Sitasuwan, Bussaratid, Ruttanaumpawan, & Chotinaiwattarakul, 2014).

Statistical analyses

Analyses were performed separately for the Chicago cohort and the Thailand cohort since different questionnaires were used to assess morningness–eveningness. Data are presented as mean (SD) or frequency (%).

Univariate linear regression analyses tested associations between CES-D scores and demographic, glycemic, sleep, and morningness–eveningness variables. A hierarchical multiple regression analysis tested whether morningness–eveningness was independently associated with depressive symptoms. Demographic, glycemic, and sleep variables significantly associated with CES-D scores in the univariate analyses (p < 0.05) were entered in the first step. In the second step, MEQ or CSM scores were entered. Cohen’s ƒ² was calculated to estimate as effect size of MEQ or CSM on CES-D scores, with effect sizes of 0.02, 0.15, and 0.35, which are termed small, medium, and large, respectively (Cohen, 1988). Finally, a logistic regression analysis tested for independent predictors for being at risk of clinical depression (CES-D scores ≥ 16). Collinearity analysis demonstrated no collinearity among the variables. Analyses were performed using SPSS version 18.0 (Chicago, IL).

Results

Chicago cohort

The demographic, diabetes, and sleep characteristics, as well as depressive symptoms and morningness–eveningness of the 182 participants are shown in Table 1. Mean age was 58.8 years, 69.8% of the participants were female, and 59.9% had poor sleep quality as indicated by PSQI score > 5. The mean (SD) CES-D score was 13.7 (9.1), and 65 (35.7%) participants scored above the cutoff for clinically significant levels of depression symptoms (≥ 16). Of those with CES-D scores ≥ 16, 19 (29.2%) were on antidepressants. Nineteen participants with scores below the cutoff for depression were using antidepressants.

Table 1. Baseline demographic, glycemic and sleep characteristics, depressive symptoms, and morningness–eveningness preference.

Variables	Chicago cohort (n = 182)	Thailand cohort (n = 251)
Age (years)	58.8 (12.9)	56.5 (11.3)
Female (n) (%)	127 (69.8)	145 (57.8)
White race (n) (%)	51 (28.0)	N/A
Night shift work (n) (%)	N/A	60 (23.9)
BMI (kg/m^2)	35.6 (8.4)	28.4 (5.0)
Diabetes duration (years)	13.7 (10.7)^a	10.9 (9.2)
HbA1c (%)	7.8 (1.6)	7.7 (1.5)
Insulin use (n) (%)	104 (57.1)	93 (37.1)
Diabetes complication ≥ 1 (n) (%)	132 (72.5)	149 (60.8)^c
PSQI score	6.8 (4.2)	8.0 (3.5)^d
PSQI score > 5 (n) (%)	109 (59.9)	187 (74.5)^d
CES-D	13.7 (9.1)	11.9 (6.4)
CES-D score ≥ 16 (n) (%)	65 (35.7)	55 (21.9)
Antidepressant use (n) (%)	38 (20.9)	3 (1.2)
MEQ score	59.4 (9.1)^b	N/A
CSM score	N/A	43.6 (6.0)

Data are presented as mean (SD) or n (%). ^an = 181, ^bn = 180, ^cn = 245, ^dn = 250.

BMI = body mass index, HbA1c = hemoglobin A1c, PSQI = Pittsburgh Sleep Quality Index, CES-D = Center for Epidemiological Studies Depression score, MEQ = Morningness–Eveningness questionnaire, CSM = Composite Score of Morningness.

Univariate regression analyses to determine associations between CES-D scores and demographic variables, diabetes variables (HbA1c, diabetes duration, insulin use, and diabetes complications), PSQI scores, and morningness–eveningness are shown in Table 2. Younger age, being female, non-White race, poorer glycemic control, insulin use, and poorer sleep quality were associated with greater depressive symptoms. In addition, eveningness (lower MEQ scores) was associated with higher CES-D scores.

Table 2. Univariate linear regression analyses between depressive symptoms (CES-D Score) and participants’ characteristics.

Variables	Chicago cohort		Thailand cohort
	B	p value	B	p value
Age (years)	–0.147	0.005	–0.079	0.027
Female	3.076	0.036	2.070	0.012
White	–4.308	0.004	N/A	N/A
Night shift work	N/A	N/A	0.864	0.070
BMI (kg/m^2)	0.050	0.538	–0.074	0.361
Diabetes duration (years)	0.067	0.293	0.009	0.841
HbA1c (%)	1.226	0.003	–0.129	0.832
Insulin use	4.471	0.001	–0.244	0.773
Diabetes complications ≥ 1	2.465	0.103	–0.684	0.420
PSQI score	1.366	< 0.001	0.830	< 0.001
MEQ score	–0.251	0.001	N/A	N/A
CSM score	N/A	N/A	–0.249	< 0.001

B = unstandardized coefficient.

HbA1c = hemoglobin A1c, PSQI = Pittsburgh Sleep Quality Index, CES-D = Center for Epidemiological Studies Depression score, MEQ = Morningness–Eveningness questionnaire, CSM = Composite Score of Morningness.

Independent association between morningness–eveningness and depressive symptoms was determined by hierarchical multiple regression controlling for relevant variables (Table 3). Demographic variables, HbA1c, insulin use, and sleep quality explained 43.6% of the variance in CESD scores. MEQ score was added in the second step. This revealed that lower MEQ score (greater eveningness) was significantly associated with higher CES-D score (unstandardized coefficient, B = –0.117, p = 0.048). This model explained an additional 1.2% of the variance in CES-D score (∆R² = 0.012, p = 0.048, total adjusted R² = .446), which indicated that MEQ score contributed significantly, though only modestly, to the model’s explanation of the variance of CES-D score above and beyond demographic variables, HbA1c, insulin use and sleep quality. The ƒ² was 0.023, indicating a small effect size.

Table 3. Hierarchical multiple linear regression with CES-D score as an outcome, Chicago cohort (N = 180).

Variables	Model 1		Model 2
	B	p value	B	p value
Age	–0.077	0.074	–0.078	0.069
Female	0.754	0.527	0.853	0.471
White	–1.262	0.305	–1.313	0.282
HbA1c	0.265	0.470	0.165	0.653
Insulin use	2.321	0.046	2.062	0.076
PSQI score	1.259	< 0.001	1.223	< 0.001
MEQ score			–0.117	0.048
Adjusted R^2	.436		.446
∆R^2	—		0.012	0.048

B = unstandardized coefficient, CES-D = Center for Epidemiological Studies Depression score, HbA1c = hemoglobin A1c, PSQI = Pittsburgh Sleep Quality Index, MEQ = Morningness–Eveningness questionnaire.

Variables significantly associated with CES-D score in the univariate regression analysis (p < 0.05) were included in model 1. MEQ score was included in model 2 to evaluate its additional contribution to CES-D score.

To determine variables associated with risk for clinical depression (CES-D score ≥16), a logistic regression analysis was performed (Table 4). This revealed that only poorer sleep quality, but not MEQ score, was significantly associated with clinical depression risk.

Table 4. Logistic regression with risk for clinical depression (CES-D score ≥ 16) as an outcome, Chicago cohort (n = 180).

Variables	B	p value	Odds ratio (95% CI)
Age	–0.025	0.099	0.975 (0.946–1.0045
Female	0.142	0.747	1.152 (0.488–2.722)
White	–0.223	0.636	0.800 (0.319–2.011)
HbA1c	0.024	0.854	1.024 (0.791–1.326)
Insulin use	0.699	0.101	2.012 (0.872–4.645)
PSQI score	0.274	< 0.001	1.325 (1.188–1.457)
MEQ score	–0.025	0.237	0.975 (0.935–1.017)

B = unstandardized coefficient, CES-D = Center for Epidemiological Studies Depression score, HbA1c = hemoglobin A1c, PSQI = Pittsburgh Sleep Quality Index, MEQ = Morningness–Eveningness Questionnaire.

Thailand cohort

Demographic, glycemic, and sleep variables, as well as depressive symptoms and morningness–eveningness of the 251 Thai participants, are shown in Table 1. Mean age was 56.5 years, 57.8% of the participants were female and 74.5% had poor sleep quality as indicated by PSQI score > 5. The mean (SD) CES-D score was 11.9 (6.4), and 55 (21.9%) participants scored above the cutoff for risk for clinical depression. The use of antidepressant was rare, as only 3 participants (1.2%) were on antidepressants, and an additional 13 (5.2%) were on anxiolytics. Of those with CES-D scores ≥ 16 (n = 55), only 2 (3.6%) were on antidepressants.

Univariate regression analyses to determine associations between CES-D scores and demographic variables, diabetes variables (HbA1c, diabetes duration, insulin use and diabetes complications), sleep quality, and morningness–eveningness are shown in Table 2. Younger age, being female, and poorer sleep quality were associated with greater depressive symptoms. In addition, eveningness (lower CSM scores) was associated with higher CES-D scores. Diabetes duration, HbA1c, insulin use, diabetic complications, and night shift work were not related to depressive symptoms.

Independent association between morningness–eveningness and depressive symptoms was determined by hierarchical multiple regression controlling for relevant variables (Table 5). Demographic variables and sleep quality explained 24.4% of the variance in CES-D scores. CSM score was added in the second step. This revealed that a lower CSM score (more eveningness) was significantly associated with higher CES-D score (unstandardized coefficient, B = –0.147, p = 0.016). Similar to the Chicago cohort, this model explained an additional 1.8% of the variance in CES-D score (∆R² = 0.018, p = 0.016, total adjusted R² = .265), which indicated that CSM score contributed significantly, though only modestly, to the model’s explanation of the variance of CES-D score above and beyond demographic variables and sleep quality. The ƒ² was 0.018, indicating a small effect size.

Table 5. Hierarchical multiple linear regression with CES-D score as an outcome, Thailand cohort (n = 250).

Variables	Model 1		Model 2
	B	p value	B	p value
Age	–0.069	0.030	–0.055	0.083
Sex	2.219	0.002	2.1194	0.002
PSQI score	0.833	< 0.001	0.789	< 0.001
CSM score			–0.147	0.016
Adjusted R^2	.239		.265
∆R^2	—		0.018	0.016

B = unstandardized coefficient, CES-D = Center for Epidemiological Studies Depression score, PSQI = Pittsburgh Sleep Quality Index, CSM = Composite Score of Morningness.

Variables significantly associated with CES-D score in the univariate regression analysis (p < 0.05) were included in model 1. CSM score was included in model 2 to evaluate its additional contribution to CES-D score.

To determine variables significantly associated with risk for clinical depression (CES-D score ≥ 16), a logistic regression analysis was performed (Table 6). This revealed that only poorer sleep quality, but not CSM score, was significantly associated with clinical depression risk.

Table 6. Logistic regression with risk for clinical depression (CES-D score ≥ 16) as an outcome, Thailand cohort (n = 250).

Variables	B	p value	Odds ratio (95% CI)
Age	–0.012	0.421	0.988 (0.961–1.017)
Sex	0.329	0.324	1.389 (0.723–2.671)
PSQI score	0.219	< 0.001	1.244 (1.130–1.371)
CSM score	–0.030	0.251	0.970 (0.921–1.022)

B = unstandardized coefficient, CES-D = Center for Epidemiological Studies Depression score, PSQI = Pittsburgh Sleep Quality Index, CSM = Composite Score of Morningness.

Discussion

In this study of type 2 diabetes patients from two different geographic locations (United States and Thailand), we demonstrated for the first time that eveningness was independently associated with higher depressive symptoms, after adjusting for demographic variables, diabetes-related factors, and sleep quality. Although direct comparisons between the two cohorts are difficult given that different questionnaires were utilized to assess morningness–eveningness, the predictive value of circadian phase preference on depressive symptoms appears to be similar between the two populations. While significant, the relatively small amount of variance in depression explained by morningness–eveningness may explain why greater eveningness did not predict clinical depression risk. In both cohorts, poorer sleep quality was also independently associated with higher depressive symptoms, consistent with previous findings (Holt, de Groot, & Golden, 2014). These results suggest that there are modest, but significant associations among eveningness, sleep quality, and depressive symptoms in patients with type 2 diabetes, regardless of geographic location. Similar to the general population, these data further support the role of the circadian timing system in psychological functioning.

The relationship between circadian regulation and depression is likely bidirectional. A disruption in the sleep wake cycle is one of the main symptoms in depressed patients. It is also known that circadian disruption such as jet lag or shift work can lead to alterations in mood, affect, or cognitive function (Karatsoreos, 2014). In patients with a depressive disorder, circadian misalignment, as measured by the phase angle difference between dim light melatonin onset and sleep timing, was correlated with depression severity (Emens, Lewy, Kinzie, Arntz, & Rough, 2009). Recent genetic studies have elucidated a link between circadian regulation and depression. In a rat model of depression, alterations in the expression of clock genes (Bmal1 and Per2) were associated with the induction of a depression-like state (Christiansen, Bouzinova, Fahrenkrug, & Wiborg, 2016). In humans, a study of twins found a significant genetic correlation between morningness–eveningness and depression, suggesting that both conditions share a significant amount of the underlying genetic variance (Toomey, Panizzon, Kremen, Franz, & Lyons, 2015). Moreover, several genetic variants of the clock gene machinery have been found to be associated with elevated risks of depression in population-based studies, including polymorphisms in Cry1 (Hua et al., 2014), Cry2 (Byrne et al., 2014) and aryl hydrocarbon receptor nuclear translocator-like (ARNTL; Rajendran & Janakarajan, 2016).

The relationship between diabetes and depression is also likely bidirectional and both share some biological and psychological origins. Environmental factors (sedentary lifestyles, obesity, low socioeconomic status) may promote activation of hypothalamic pituitary axis and increase an inflammatory state with elevated proinflammatory cytokines (C-reactive protein, TNF-α) in susceptible individuals, leading to both disorders in parallel (Moulton et al., 2015). In patients with diabetes, the additional burden of diabetes management and hypo- or hyperglycemia may contribute to the risk of depression and alterations in brain functions (Holt et al., 2014). Common lifestyle factors in those with depression, such as unhealthy diet, being sedentary, and the use of certain antidepressants, may pose a risk for diabetes development (Barnard, Peveler, & Holt, 2013; Holt et al., 2014). The role of circadian regulation in the pathogenesis of both diabetes and depression should be considered. It is well known that the circadian system plays a significant role in metabolism. Experimental circadian misalignment in healthy volunteers resulted in impaired glucose tolerance (Scheer, Hilton, Mantzoros, & Shea, 2009). Longitudinal studies also revealed that shift work is associated with increased risk of developing diabetes (Anothaisintawee, Reutrakul, Van Cauter, & Thakkinstian, 2015), and type 2 diabetes patients with later chronotype (more evening preference) exhibit poorer glycemic control (Reutrakul et al., 2013). Moreover, certain genotypes of clock and Bmal1 were associated with evening preference, resistance to weight loss, metabolic syndrome, and susceptibility to type 2 diabetes (Garaulet et al., 2012; Scott, Carter, & Grant, 2008; Woon et al., 2007). Some have postulated that dysregulation of the circadian clock plays a role in metabolic comorbidity in psychiatric disorders (Barandas, Landgraf, McCarthy, & Welsh, 2015). Adverse environmental factors may precipitate disturbances of circadian rhythms, leading to increased risk for both diabetes and depression. While certain polymorphisms of clock genes were associated with either abnormal glucose metabolism or depression, studies in mice revealed that some polymorphisms demonstrated both metabolic and mood-related behavior phenotypes (Barandas et al., 2015; Dallmann, Touma, Palme, Albrecht, & Steinlechner, 2006). Further investigation in humans is warranted.

Our findings confirm the high prevalence of depression in diabetes patients in both countries, as previously described (Anderson et al., 2001; Thongsai et al., 2013). The rate of antidepressant use in the Chicago participants (20.9%) appeared to be higher than that reported in the general U.S. population (11%; Pratt, Brody, & Qiuping, 2011). The rate of antidepressant use in Thailand cohort, however, is quite low, suggesting that depression may have not been recognized by the participant’s physicians or that antidepressants were not a preferred treatment for some reason. Despite the differences in geographic locations and depression recognition, eveningness contributed to greater depressive symptoms in both cohorts. Treatment of depression in patients with diabetes with either cognitive behavioral therapy or medication can result in improvements in depressive symptoms and increased quality of life, especially when combined with diabetes education and exercise, although the effects on glycemic control have been mixed (Markowitz, Gonzalez, Wilkinson, & Safren, 2011; Young-Hyman et al., 2016). In patients with depression, chronotherapeutic interventions such as bright light therapy and sleep phase advance have been shown to be effective (Terman, 2007; Wu et al., 2009). Whether there is an interaction among morningness–eveningness, the degree of phase advancement, and the effectiveness of therapy remains unknown. In a study of patients with seasonal affective disorder (SAD), fixed-time light therapy revealed that there was no relationship between morningness–eveningness and therapy success (Knapen, Gordijn, & Meesters, 2016). Similarly, in another study of patients with SAD, both fluoxetine and light therapy exerted antidepressant effects and resulted in a phase advance. However, the degree of symptom change did not correlate with the degree of phase change (Murray et al., 2005). More recently, a study of participants with insomnia using behavioral treatment (sleep schedule, stimulus control, and adherence promotion) resulted in a shift toward morningness, and the degree of the shift correlated with the reduction in depressive symptoms (Hasler, Buysse, & Germain, 2016). In another study of insomnia patients, cognitive behavioral therapy resulted in sleep improvement in all chronotypes, but those with greater eveningness also experienced a reduction in depressive symptom severity (Bei, Ong, Rajaratnam, & Manber, 2015). Whether chronotherapeutic interventions will result in improvement in depressive symptoms (and possibly metabolic control) in diabetes patients, especially those with more evening preference, should be investigated.

Agomelatine is a melatonergic antidepressant, which has been shown to produce antidepressant effects, along with an increase in the relative amplitude of the circadian rest–activity cycle and improvements in sleep (Kasper et al., 2010). In a recent randomized controlled study of 116 type 2 diabetes patients, agomelatine for 12 weeks, compared to paroxetine, resulted in lower depression scores (Kang et al., 2015). Interestingly, the agomelatine group had a significant reduction in their HbA1c levels compared to those in the paroxetine group. It is possible that by manipulating circadian rhythmicity, the medication targets the pathogenesis of both disorders and results in improvements of both metabolism and depressive symptoms. This should be explored in future research. Whether diabetes patients with evening preference will receive more benefit from this treatment than those with morning preference remains to be investigated.

A strength of our study is its inclusion of exclusively type 2 diabetes patients from two different countries. However, there are limitations. The cross-sectional design did not allow us to determine the direction of causality in the associations between morningness–eveningness and depression. Antidepressant can impact the circadian system and is associated with new-onset diabetes (Bhattacharjee, Bhattacharya, Kelley, & Sambamoorthi, 2013; Jones & Benca, 2015); thus, antidepressants could play a role in the interactions between eveningness and depressive symptoms in patients with type 2 diabetes. We do not think antidepressants accounted entirely for this association, however, since very few patients in the Thai cohort were prescribed medication for depression symptoms. Nevertheless, future research may need to explore the temporal associations between the development of diabetes and the onset of depression in order to better understand these relationships. In addition, as the mean age of our population was the mid-50s (likely due to type 2 diabetes being more common in an older population), whether the findings will be reproducible in younger patients remains to be investigated. However, the association between eveningness and depressive symptoms has been reported across age groups, including healthy young adults (Jankowski, 2016; Merikanto et al., 2013). Lastly, the effect size of morningness–eveningness on depressive symptoms was relatively small, suggesting that other factors also play a role in depressive symptoms in this patient group.

In summary, eveningness is independently associated with greater depressive symptoms in patients with type 2 diabetes in both the United States and Thailand. Future research should explore whether chronotherapeutic interventions will result in improvement in depressive symptoms, and possibly metabolic control, in this patient group.

Funding

This study was funded by a departmental grant, section of Endocrinology, Rush University Medical Center, Chicago, Illinois; Mahidol University, Bangkok, Thailand; and The Endocrine Society of Thailand.

Additional information

Funding

This study was funded by a departmental grant, section of Endocrinology, Rush University Medical Center, Chicago, Illinois; Mahidol University, Bangkok, Thailand; and The Endocrine Society of Thailand.