Association between depressive symptoms and metabolic syndrome is not explained by antidepressant medication: Results from the PPP-Botnia Study

Abstract

Introduction. To study whether the frequently reported association between depressive symptoms and the metabolic syndrome (MetS) and its individual components are secondary to the use of antidepressant medication and to established diabetes or cardiovascular diseases (CVD).

Patients and methods. A population-based, random sample of 4,967 women and men aged 18–75 years. MetS was defined according to the new, harmonized criteria. Glucose tolerance was assessed by oral glucose tolerance test (OGTT). CVD, depressive symptoms, and use of antidepressant medication were self-reported.

Results. The odds for having the MetS increased over 10%for each standard deviation increase in depressive symptoms. Users of antidepressant medication had more than 50% increased odds for having the MetS. Depressive symptoms were also associated with higher glucose response during the OGTT, higher serum triglyceride and lower HDL-cholesterol concentrations, and higher waist circumference, while use of antidepressant medication was associated with higher triglycerides, waist circumference, and systolic blood pressure. The associations of depressive symptoms were not secondary to use of antidepressant medication and were not explained by established diabetes or CVD.

Discussion. Depressive symptoms, the MetS, and the individual components of MetS are related. These associations are not driven by use of antidepressant medication, established diabetes, or CVD.

Key words::

• Antidepressive agents
• depression
• metabolic syndrome

Abbreviations
CVD	=	cardiovascular disease
DBP	=	diastolic blood pressure
HPAA	=	hypothalamic-pituitary-adrenal axis activity
IGT	=	impaired glucose tolerance
MET	=	metabolic equivalent
MetS	=	metabolic syndrome
OGTT	=	oral glucose tolerance test
SBP	=	systolic blood pressure
SNRI	=	serotonin-norepinephrine reuptake inhibitor
SSRI	=	selective serotonin reuptake inhibitor

Key messages

• The frequently reported association between depressive symptoms and the metabolic syndrome is not secondary to use of antidepressant medication, or established diabetes or cardiovascular diseases.

Introduction

Mounting empirical evidence suggests that depression and subclinical depressive symptoms are associated with the metabolic syndrome (MetS) (1–6) and type 2 diabetes (7,8). MetS is a cluster of risk factors composed of insulin resistance, obesity, dyslipidemia, and elevated blood pressure (9) that predicts future risk for cardiovascular diseases and type 2 diabetes (10,11). A recent study suggests that the use of antidepressant medication is a stronger predictor of type 2 diabetes than are depressive symptoms (12). As use of antidepressant medication and depressive symptoms are strongly related (2,3,12), and since some antidepressants may per se contribute to weight gain (13) and to hyperglycemia in diabetic patients (14), an important question is whether an association between depressive symptoms and the MetS is secondary to the use of antidepressant medication. Here we report the associations between depressive symptoms, use of antidepressant medication, and the MetS in a large Finnish population-based study, the PPP-Botnia Study (PPP = prevalence, prediction, and prevention of diabetes) (15,16).

Recently, the American Diabetes Association jointly with the European Association for the Study of Diabetes (17) and the World Health Organization expert consultation (18) have provided critical re-evaluations on the definition of the MetS and its utility as a clinical concept: the value of including established diabetes and cardiovascular disease (CVD) in the definition is questionable, the rationale for cut-off values used in determining the risk factor levels are ill-defined, and the disease risk associated with the clustering of the risk factors may not be greater than that associated with each of its individual components. Therefore, the present study also reports associations between depressive symptoms, use of antidepressant medication, and the MetS by excluding individuals with established diabetes and CVD and testing the associations with the individual components of the MetS as continuous, instead of using cut-off points for components as presented in the clinical criteria for the MetS.

Patients and methods

Participants

The PPP-Botnia Study is a population-based study in the Botnia region of western Finland (15,16). The study is designed to obtain accurate estimates of prevalence and risk factors for diabetes, impaired glucose regulation, and the MetS in the population aged 18–75 years and to use this information for prediction and prevention of diabetes. The current study was initiated in 2004 in five Botnia centers (Närpes, Malax-Korsnäs, Korsholm, Vasa, and Jakobstad) comprising a total population of 135,000 persons. Using the population registry we selected an age-adjusted random sample of subjects aged 18–75 years (96,000 subjects) representing on average 9% of the population. Of the 9,518 invited individuals 5,208 (2,443 men and 2,765 women) (55%) participated. Of the participants, 4,967 (2,336 men and 2,631 women) had complete data available on the components of the MetS, the use of antidepressant medication, and depressive symptoms, and did not use antipsychotic medication (n = 36) or lithium (n = 5). Those with incomplete data (n = 200) were older, more frequently retired, consumed less alcohol, had higher glucose concentrations (fasting and 2-hour glucose) and higher systolic blood pressure (SBP) levels, larger waist circumferences, and reported more frequently a family history of diabetes compared with those with complete data. The participants gave their written informed consent, and the study protocol was approved by the Ethics Committee of Helsinki University Central Hospital, Finland.

Metabolic syndrome

Body-weight and height were measured with subjects in light indoor clothing and without shoes. Waist circumference was measured with a soft tape, on standing subjects, midway between the lowest rib and the iliac crest. Two blood pressure recordings were obtained from the right arm of a sitting subject after 10 min of rest, and the mean value was calculated.

The subjects participated in an oral glucose tolerance test by ingesting 75 g of glucose after a 12-h overnight fast. Subjects with known diabetes and on antidiabetic medication or with fasting plasma glucose >10 mmol/L did not take part in the oral glucose tolerance test (OGTT) (n = 19). During the OGTT, samples for plasma glucose and insulin were drawn at 0 min, 30 min, and 120 min. Diagnosis of diabetes was based on the results from the OGTT or a history of previously known diabetes. The diagnosis of diabetes was based on the WHO criteria (19). Thus, subjects with a fasting plasma glucose ≥ 7.0 mmol/L and/or 2-h plasma glucose ≥11.0 during an OGTT were considered to have diabetes. Fasting blood samples were drawn for the measurement of serum total cholesterol, HDL cholesterol, and triglycerides.

According to the harmonized clinical criteria (9), the definition of the MetS requires that at least three of the following five criteria have to be met: fasting plasma glucose (≥ 5.6 mmol/L and/or diabetes medication), triglycerides (≥ 1.7 mmol/L and/or drug treatment of high triglycerides), HDL cholesterol (< 1.0 mmol/L in men and < 1.3 mmol/L in women and/or drug treatment for low HDL cholesterol), blood pressure (≥ 130/85 mmHg and/or use of antihypertensive medication), and waist circumference (ethnicity- and country-specific cut-off; we used the European Cardiovascular Society's cut-offs, ≥ 102 cm for men and ≥ 88 cm for women (20)).

Assays

Plasma glucose was measured with a glucose dehydrogenase method (HemoCue, Ängelholm, Sweden) and serum insulin by a fluoroimmunoassay (Delphia; Perkin-Elmer Finland, Turku, Finland). Serum triglyceride and HDL-cholesterol concentrations were analyzed with an enzymatic method on a Konelab 60i analyser (Thermo Electron Oy, Vantaa, Finland).

Depressive symptoms and use of antidepressive medication

Depressive symptoms were self-rated using the SF-36/RAND-36 (21). The Mental Health Index, a subscale of SF-36, is used to capture four major dimensions of mental health: anxiety, depression, loss of behavioral/emotional control, and psychological well-being (22), and the scale has been shown to have high sensitivity and specificity for detecting clinical depression (23), and it has been used to measure specifically depressive symptoms (7). However, a Finnish validation study of SF-36 found that the Vitality scale items correlate positively with the Mental Health Index and concluded that the Vitality scale items are also important in capturing depression in a Finnish population (24). Therefore, items of the Mental Health Index and the Vitality scale were summed (after reverse scoring the positive items) to measure depressive symptoms on a scale asking participants how much of the time during the preceding 4 weeks they felt full of life, nervous, happy, calm and peaceful, worn out, full of energy, downhearted and blue, tired, or so down nothing could cheer them up. The items were rated on a six-point scale (not at all, on very few occasions, some of the time, good part of the time, most of the time, all of the time). The reliability for these items was high (Cronbach's α = 0.89), and a factor analysis (using an eigenvalue criterion of one) of the items supported a unidimensional, single-scale solution (data not shown).

The use of antidepressant medication was self-reported in a questionnaire and confirmed in an interview by a research nurse specifying the type of medication they used. We considered current use of selective serotonin reuptake inhibitor (SSRI)/ serotonin-norepinephrine reuptake inhibitor (SNRI) medication (n = 116), tricyclic medication (n = 43), tricyclic and SSRI/SNRI (n = 11) as an indicator of use of antidepressant medication. In addition, information on four participants’ type of medication was insufficient. In total 174 participants, 132 women and 42 men, reported using one or more products of antidepressant medication.

Diabetes and CVD

Diagnosis of diabetes was based on the results from the OGTT or a history of previously diagnosed diabetes. In uncertain cases, the diagnosis was confirmed from patient records. Participants filled in a standardized health questionnaire covering information regarding a history of coronary heart disease (including myocardial infarction, coronary bypass surgery, and percutaneous coronary interventions), stroke (including both ischemic and hemorrhagic stroke), and invasive procedures for peripheral arterial disease. Cardiovascular disease was considered present if any of these manifestations were present. Those with established diabetes or CVD (n = 539), in comparison to the other participants with no such diseases, were older, more frequently men and retired, had the MetS more frequently, had a higher fasting glucose, 2-h glucose, waist circumference, SBP, diastolic blood pressure (DBP), triglycerides, and lower HDL cholesterol, consumed less alcohol, reported more frequently a family history of diabetes, and had higher depressive symptoms scores (all P values < 0.008; data not shown).

Mediating and confounding factors

The subjects self-reported their weekly alcohol consumption (g/week), current smoking status (yes versus no or former smoker), occupational status (categorized according to the classification system of Statistics Finland: manual workers, junior clericals, senior clericals, students, and retirees), and family history of known diabetes (yes versus no) in at least one first-degree relative (father, mother or sibling). In addition, frequency and intensity of current physical activity and physical activity during the past 12 months were assessed using the validated Kuopio Ischemic Heart Disease Questionnaire (25). This questionnaire provides detailed information on common life-style, commuting, and leisuretime physical activity and enables assessment of total physical activity as metabolic equivalent (MET) hours per week (MET × hours/week). Based upon leisuretime activity, the participants were assigned into two groups: if they performed more than 30 minutes physical activity three or more times per week with an intensity resulting in breathlessness and/or sweating they were assigned to regularly exercising group (yes), and if they performed less or performed no activity they were assigned to less/no exercising group (no).

Statistical analyses

Logistic regression analyses, odds ratios (OR) and 95% confidence intervals (CI) were computed to examine if depressive symptoms and antidepressant medication use were associated with the MetS. Multiple linear regression analyses, unstandardized regression coefficients, and 95% CI were computed to examine associations between depressive symptoms and antidepressant medication use with fasting and 2-h plasma glucose, triglycerides, HDL cholesterol, waist circumference, and SBP and DBP. For those analyses that tested if any potential associations between depressive symptoms with the MetS and its individual components were secondary to antidepressant medication use, individuals reporting antidepressant medication use were excluded and the analyses were re-run.

Variables were log-transformed where appropriate. The associations were adjusted for mediating and confounding factors. The associations were re-run after excluding participants with established diabetes and CVD.

Finally, because associations between depressive symptoms, use of antidepressant medication, and the MetS may be moderated by sex (1,3,4,6), and since our own data demonstrated that women scored higher on depressive symptoms (P< 0.01) and reported more frequent use of antidepressant medication (P< 0.01), we tested if any of the associations varied for men and women by including an interaction term, ‘sex × depressive symptoms/antidepressant medication use’ in the models. In no instance was there a significant sex interaction term (P values > 0.38; data not shown). For this reason we report the results in both sexes combined.

Results

Characteristics of the study sample are presented in Table I according to the harmonized MetS clinical criteria. Depressive symptoms were associated with use of antidepressant medication: per each one standard deviation (SD) unit increase in depressive symptoms, the odds for antidepressant medication use increased 2.62-fold (95% CI 2.24–3.07, P = 0.001; for men OR 2.89, 95% CI 2.11–3.96, P = 0.001; for women OR 2.53, 95% CI 2.11–3.04, P = 0.001).

Table I. Characteristics of the sample according to the harmonized clinical criteria of the metabolic syndrome (MetS).

	MetS
Characteristic	No n = 3,586 (72.2%)	Yes n = 1,381 (27.8%)	P
Sex, men, n (%)	1,641 (45.8)	695 (50.3)	0.004
Age (years)	46.1 ± 15.4	56.6 ± 13.2	0.001
Fasting glucose (mmol/L)	5.2 ± 0.8	5.9 ± 1.1	0.001
120 minute glucose (mmol/L)	5.0 ± 1.6	6.8 ± 2.9	0.001
New diabetes based on OGTT, n(%)	32 (0.9)	78 (5.6)	0.001
Previously known diabetes, n (%)	31 (0.9)	140 (10.1)	0.001
Established CVD, n (%)	176 (4.9)	239 (17.3)	0.001
Waist circumference (cm)
Men	91.0 ± 8.8	105.0 ± 10.4	0.001
Women	79.5 ± 9.5	96.6 ± 10.9	0.001
Triglycerides (mmol/L)	1.1 ± 0.5	1.9 ± 1.1	0.001
HDL cholesterol (mmol/L)
Men	1.33 ± 0.33	1.07 ± 0.31	0.001
Women	1.63 ± 0.37	1.30 ± 0.37	0.001
SBP (mmHg)	130 ± 18	144 ± 19	0.001
DBP (mmHg)	79 ± 10	86 ± 10	0.001
Current smoker, n (%)	507 (14.2)	218 (16.1)	0.098
Regular exercise, n (%)	2,017 (57.2)	652 (48.5)	0.001
Alcohol consumption, n (%)
None	829 (23.9)	424 (31.9)	0.001
12–48 g/week	1,654 (47.6)	568 (42.8)
≥ 60 g/week	989 (28.5)	336 (25.3)
Occupational status, n (%)
Senior clerical	655 (18.7)	169 (12.4)	0.001
Junior clerical	1,096 (31.4)	335 (24.6)
Manual workers	890 (25.5)	344 (25.3)
Retired	645 (18.5)	506 (37.2)
Students	209 (6.0)	8 (0.6)
First-degree family history for diabetes, n (%)	932 (26.0)	549 (39.8)	0.001

Self-reported alcohol dose per week was converted into g/week.

Depressive symptoms, use of antidepressant medication, and the MetS

Depressive symptoms were significantly associated with increased odds for having the MetS (Table II). Depressive symptoms remained a significant predictor of the MetS in all models except for one: when participants with established diabetes and CVD were excluded and mediating and confounding factors were controlled, the association was rendered non-significant (Table II). When individuals with depressive symptoms scores in the top and the lower three quartiles were contrasted the results remained virtually identical (Table II).

Table II. Prevalence of the metabolic syndrome according to presence of depressive symptoms and use of antidepressant medication.

	Total sample			Subjects without diabetes or CVD^a
Depression	n	Odds ratio (95% CI)	P	n	Odds ratio (95% CI)	P
Depressive symptoms (SD units)
Sex and age-adjusted model	4,967	1.11 (1.04–1.19)	0.002	4,428	1.10 (1.02–1.18)	0.013
Fully adjusted model	4,967	1.10 (1.03–1.18)	0.006	4,428	1.07 (1.00–1.16)	0.067
Participants with antidepressant medication use excluded^a,b	4,793	1.10 (1.03–1.17)	0.006	4,280	1.09 (1.01–1.17)	0.026
Depressive symptoms (top quartile versus lower three quartiles)
Sex and age-adjusted model	4,967	1.28 (1.11–1.49)	0.001	4,428	1.24 (1.05–1.46)	0.010
Fully adjusted model	4,967	1.26 (1.07–1.47)	0.004	4,428	1.19 (1.00–1.41)	0.053
Participants with antidepressant medication use excluded^a,b	4,793	1.28 (1.10–1.49)	0.002	4,280	1.24 (1.05–1.47)	0.011
Antidepressant medication use (no versus yes)
Sex and age-adjusted model	4,967	1.51 (1.08–2.10)	0.015	4,428	1.51 (1.05–2.18)	0.027
Fully adjusted model	4,967	1.52 (1.07–2.15)	0.019	4,428	1.46 (0.99–2.14)	0.054

The metabolic syndrome was defined using the harmonized clinical criteria. Fully adjusted refers to a model adjusting for sex, age, physical activity, alcohol consumption, current smoking status, occupational status, and family history for diabetes.

^aNumber of participants excluded was 539.

^bModel was adjusted for sex and age.

Antidepressant medication use was also significantly associated with increased odds for having the MetS (Table II). However, after exclusion of participants with established diabetes and CVD and controlling for mediating and confounding factors the association was no longer significant.

Depressive symptoms, use of antidepressant medication, and individual components of the MetS

Table III and Table IV present the associations between depressive symptoms, use of antidepressant medication, and individual components of the MetS as continuous variables. Because in rare instances the associations lost statistical significance, when individuals with established diabetes and CVD were excluded, only the analyses excluding these individuals are presented. Depressive symptoms were significantly associated with higher 2-h glucose, higher triglycerides, and lower HDL-cholesterol concentration (Table III), and higher waist circumference (Table IV). These associations remained significant in models adjusting for mediating and confounding factors, and when excluding those participants who reported antidepressant medication use. Depressive symptoms were not associated with fasting glucose (Table III), SBP, or DBP (Table IV). However, depressive symptoms were associated with higher fasting glucose when participants with CVD or diabetes were included in the analysis (footnote of Table III). When individuals with depressive symptoms scores in the top and the lower three quartiles were contrasted the results remained virtually identical (Tables III and IV).

Table III. Associations between depressive symptoms and use of antidepressant medication with fasting and 2-h glucose, triglycerides and HDL-cholesterol. Participants with established diabetes or cardiovascular disease are excluded (n = 539).

		Fasting glucose		2-Hour glucose		Triglycerides		HDL-cholesterol
Depression	n	% change (95% CI)	P	% change (95% CI)	P	% change (95% CI)	P	% change (95% CI)	P
Depressive symptoms (SD units)
Sex and age-adjusted model	4,428	0.07 (−0.27–0.41)	0.688^a	1.45 (0.53–2.37)	0.002	3.49 (2.06–4.92)	0.001	−1.22 (−2.00–−0.44)	0.002
Fully adjustedmodel	4,428	−0.03 (−0.38–0.33)	0.880^b	1.00 (0.04–1.97)	0.041	2.50 (1.02–3.99)	0.001	−0.84 (−1.65–−0.03)	0.041
Participants with antidepressant medication use excluded^h	4,280	−0.01 (−0.36–0.33)	0.937^c	1.49 (0.53–2.44)	0.002	3.17 (1.69–4.64)	0.001	−1.05 (−1.86–−0.24)	0.011
Depressive symptoms (top quartile versus lower three quartiles)
Sex and age-adjusted model	4,428	0.47 (−0.29–1.24)	0.222^d	4.14 (2.06–6.23)	0.001	7.37 (4.14–10.60)	0.001	−2.89 (−4.66–−1.13)	0.001
Fully adjusted model	4,428	0.43 (−0.37–1.23)	0.290^e	3.84 (1.67–6.00)	0.001	5.48 (2.13–8.82)	0.001	−2.07 (−3.89–−0.26)	0.025
Participants with antidepressant medication use excluded^h	4,280	0.41 (−0.37–1.20)	0.302^f	4.39 (2.24–6.54)	0.001	6.56 (3.24–9.89)	0.001	−2.65 (−4.47–−0.82)	0.004
Antidepressant medication use (no versus yes)
Sex and age-adjusted model	4,428	1.23 (−0.62–3.07)	0.192	0.59 (−4.44–5.63)	0.818	10.08 (2.26–17.90)	0.012	−2.96 (−7.25–1.32)	0.175^g
Fully adjusted model	4,428	1.28 (−0.63–3.19)	0.190	−0.05 (−5.22–5.13)	0.984	10.51 (2.50–18.51)	0.010	−2.05 (−6.38–2.30)	0.357

Fully adjusted refers to a model adjusting for sex, age, physical activity, alcohol consumption, current smoking status, occupational status, and family history for diabetes. Fasting glucose, 2-hour glucose, triglycerides,and HDLcholesterol values were log-transformed.

Association is statistically significant (P < 0.05) when participants with established diabetes or CVD are included: ^a0.52 (0.13–0.90), P = 0.008; ^b0.40 (0.04–0.80), P = 0.048; ^c0.41 (0.01–0.81), P = 0.044; ^d1.36 (0.49–2.23), P = 0.002; ^e1.27 (0.37–2.17), P = 0.006; ^f1.26 (0.36–2.16), P = 0.006; ^g−4.07 (−8.06– −0.08), P = 0.046.

^hModel was adjusted for sex and age.

Table IV. Associations between depressive symptoms and use of antidepressant medication with waist circumference and SBP and DBP. Participants with established diabetes or cardiovascular disease are excluded (n = 539).

		Waist circumference		SBP		DBP
Depression	n	Change in cm (95% CI)	P	Change in mmHg (95% CI)	P	Change in mmHg (95% CI)	P
Depressive symptoms (SD units)
Sex and age-adjusted model	4,428	0.76 (0.44–1.09)	0.001	−0.28 (−0.76–0.21)	0.265	0.05 (−0.24–0.34)	0.748
Fully adjusted model	4,428	0.55 (0.21–0.89)	0.001	−0.24 (−0.74–0.26)	0.347	−0.05 (−0.31–0.30)	0.974
Participants with antidepressant medication use excluded^d	4,280	0.67 (0.35–1.02)	0.001	−0.17 (−0.68–0.33)	0.497	0.08 (−0.29–0.31)	0.960
Depressive symptoms (top quartile versus lower three quartiles)
Sex and age-adjusted model	4,428	1.58 (0.83–2.32)	0.001	−1.06 (−2.16–0.03)	0.058^a	−0.24 (−0.89–0.41)	0.472
Fully adjusted model	4,428	1.17 (0.41–1.94)	0.003	−0.93 (−2.06–0.20)	0.108	−0.17 (−0.85–0.51)	0.625
Participants with antidepressant medication use excluded^d	4,280	1.52 (0.76–2.27)	0.001	−0.93 (−2.06–0.21)	0.108	−0.37 (−1.05–0.30)	0.278
Antidepressant medication use (no versus yes)
Sex and age-adjusted model	4,428	3.57 (1.77–5.36)	0.001	−2.93 (−5.58–−0.28)	0.030^c	1.25 (−0.33–2.83)	0.121
Fully adjusted model	4,428	3.72 (1.89–5.55)	0.001	−2.70 (−5.40–−0.00)	0.050	1.57 (−0.06–3.19)	0.059^b

Fully adjusted refers to a model adjusting for age, sex, physical activity, alcohol consumption, current smoking status, occupational status, and family history for diabetes.

Association is statistically significant (P < 0.05) when participants with established diabetes or CVD are included:^a−1.11 (2.16– −0.58), P = 0.039; ^b1.69 (0.18–3.20), P = 0.028.

^cAssociation is statistically not significant (P > 0.05) when participants with established diabetes or CVD are included: −2.50 (−5.01–0.16), P = 0.052.

^dModel was adjusted for sex and age.

Use of antidepressant medication was significantly associated with higher triglycerides, waist circumference, and SBP, but it was unrelated to fasting, 2-h glucose, HDL-cholesterol, and DBP (Tables III and IV).

Discussion

The current study examined associations between depressive symptoms and use of antidepressant medication with the MetS and its individual components. Our findings showed that per each SD unit increase in depressive symptoms the odds for having the MetS increased over 10%. Users of antidepressant medication had more than 50% increased odds for having the MetS. These findings are, thus, in agreement with the majority of previously published studies showing an association between depression and the MetS (1–6). Our findings, however, add significantly to the existing literature by showing that even though depressive symptoms and antidepressant medication use were in our study and, as reported by others, strongly related (2,3,12,26), the association between depressive symptoms and the MetS was not secondary to the use of antidepressant medication.

With regard to our findings pertaining to the individual components of the MetS, depressive symptoms were associated with higher 2-hour glucose levels, higher triglycerides and lower HDL-cholesterol levels, and higher waist circumference, but they were not associated with fasting glucose or blood pressure values. Again the association between depressive symptoms and glucose, lipids, and waist circumference was not secondary to use of antidepressant medication. Use of antidepressant medication, in turn, was associated with higher triglycerides, waist circumference, and SBP, but not with fasting or 2-hour glucose levels, HDL-cholesterol, and DBP. The concept and the definition of the MetS have recently been criticized (17,18). The cut-off values of the components used in the definition of the MetS are arbitrary, as many of the risk factors are linearly associated with CVD (17,18). Our findings, thus, suggest that the association between depression and the MetS is not driven by the somewhat arbitrary cut-off levels used in the definition of the MetS. Rather, depression and components of the MetS seem to associate with each other in a dose-response manner. The somewhat different pattern of associations found for depressive symptoms and antidepressant medication may suggest that depressive symptoms and antidepressant medication use may contribute to the MetS and associated disorders via somewhat different mechanisms. This is not surprising given that not all individuals scoring highly on depressive symptoms are currently using antidepressant medication, and not all antidepressant medication users score highly on depressive symptoms. In other words, the self-rated depressive symptoms screen but do not diagnose individuals with clinical depression.

Further, it has been claimed that the MetS is not a syndrome but a premorbid state, and therefore individuals with established diabetes and CVD should be excluded from the definition. We excluded individuals, roughly 10% of the sample, with established diabetes and CVD from the analyses and found that even when individuals with these disorders were excluded, depressive symptoms and antidepressant medication use remained significantly associated with the MetS and with higher levels of its individual components. Our findings thus suggest that the previously established link between depression and the MetS is not driven by having diabetes or CVD.

The pathophysiological mechanisms underlying the associations cannot be addressed in the current study. Our findings may mirror alterations in the physiological stress-regulatory mechanisms. Altered activity of the hypothalamic-pituitary-adrenocortical axis (HPAA) is among the most consistently demonstrated biological abnormalities in depression (27). Higher levels of cortisol are shown to be related to the MetS (28,29). Other possible mechanisms include autonomic nervous system (ANS), inflammatory, and neurotransmitter system activity (28–31), while mechanisms related to life-style may also be involved. While our findings were not, in general, affected by controlling for smoking, alcohol consumption, physical activity, or occupational status, used as a proxy of socio-economic position, in participants without CVD or diabetes, these life-style factors rendered the association of depressive symptoms and the MetS non-significant. Life-style factors may, thus, partially explain this association in healthier populations, though this was only true for the MetS, not for its individual components. Genetic mechanisms may be involved as well. While specific genes cannot be nominated, these may relate to glucocorticoids, catecholamines, and inflammatory pathways, or to yet unknown mechanisms. Finally, an interesting further possibility relates to prenatal programming. A suboptimal early life environment reflected in smaller body size at birth has been linked with both depression (32), impaired glucose tolerance (IGT), and type 2 diabetes (33), suggesting that early life adversities may be a common underlying mechanism.

Strengths of this study lie in the population-based study design and detailed clinical examination. Also, our study offered sufficient power to test sex specificity of the associations. While the association between depression and MetS has been reported to be different in men than in women (1,3,4,6), our data did not reveal any such differences—a finding similar to a population-based study in central Finland (34).

Our study has some limitations. An obvious limitation relates to the cross-sectional study design, precluding causal inferences: while depressive symptoms may increase the odds for the MetS (1–6), there are studies pointing at a reverse possibility (34–36). A further limitation relates to the generalizability of our findings beyond Caucasians. While we did exclude participants on antipsychotic medication and lithium from the analytic sample, we do not have data available on specific psychiatric disorders, and therefore we cannot address their role in explaining our findings in any further detail. Our study does not include information on the duration of use of antidepressant medication, and therefore we cannot address the question if medication used for a longer period in time is more detrimental than the use of the medication that started more recently. Finally, 3.9% of our population (3.7% without established diabetes and CVD) reported using antidepressant medication. The prevalence using antidepressant medication is somewhat lower than that reported in some other studies (3,12,26). The potentially healthier mental state of our sample or the lower sensitivity of the self-reported measure to capture antidepressant medication use may have diminished rather than increased our ability to detect significant associations, however.

To conclude, depressive symptoms and use of antidepressant medication are associated with the MetS and its individual components. The metabolic associations of depressive symptoms were not secondary to use of antidepressant medication. None of the associations were explained by established diabetes and CVD. Our findings may, thus, inform of the premorbid mechanisms of cardio-metabolic disorders. Interventions targeted at reducing depressive symptoms may be beneficial in a long-term health promotion.

Acknowledgements

The skillful assistance of the Botnia Study Group is gratefully acknowledged.

Declaration of interest: The authors (Katri Räikkönen, Antti-Jussi Pyykkönen, Tiinamaija Tuomi, and Bo Isomaa) declare no conflict of interest. Leif Groop has been a consultant for and served on advisory boards for Sanofi-aventis, GlaxoSmithKline, Novartis, Merck, Tethys Bioscience, and Xoma and received lecture fees from Eli Lilly and Novartis. Johan G. Eriksson has been a consultant for and served on advisory boards for MSD, Bristol-Myers Squibb, AstraZeneca, Eli Lilly, and NovoNordisk.

The PPP-Botnia study has been financially supported by grants from the European Science Foundation (EuroSTRESS), the Finnish Academy, the Sigrid Juselius Foundation, Folkhälsan Research Foundation, Nordic Center of Excellence in Disease Genetics, Signe and Ane Gyllenberg Foundation, Swedish Cultural Foundation in Finland, Finnish Diabetes Research Foundation, Foundation for Life and Health in Finland, Finnish Medical Society, the Finnish Ministry of Education, Paavo Nurmi Foundation, Perklén Foundation, Ollqvist Foundation, and Närpes Health Care Foundation. The study has also been supported by the Municipal Health Care Center and Hospital in Jakobstad, Health Care Centers in Vasa, Närpes, and Korsholm.