UEF//eRepository
DSpace https://erepo.uef.fi
Rinnakkaistallenteet Filosofinen tiedekunta
2021
Comparison of the level of allostatic load between patients with major
depression and the general population
Honkalampi, Kirsi
Elsevier BV
Tieteelliset aikakauslehtiartikkelit
© 2021 The Authors
CC BY http://creativecommons.org/licenses/by/4.0/
http://dx.doi.org/10.1016/j.jpsychores.2021.110389
https://erepo.uef.fi/handle/123456789/24781
Downloaded from University of Eastern Finland's eRepository
Journal of Psychosomatic Research 143 (2021) 110389
Available online 15 February 2021
0022-3999/© 2021 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Comparison of the level of allostatic load between patients with major depression and the general population
Kirsi Honkalampi
a,*, Marianna Virtanen
a, Taina Hintsa
a, Anu Ruusunen
b,c,d,
Pekka M ¨ antyselk ¨ a
c,e, Toni Ali-Sisto
c, Olli K ¨ arkk ¨ ainen
f, Heli Koivumaa-Honkanen
b,g, Minna Valkonen-Korhonen
b,g, Georgia Panayiotou
h, Soili M. Lehto
i,j,kaSchool of Educational Sciences and Psychology, University of Eastern Finland, Joensuu, Finland
bDepartment of Psychiatry, Kuopio University Hospital, Kuopio, Finland
cInstitute of Public Health and Clinical Nutrition, University of Eastern Finland, Kuopio, Finland
dDeakin University, iMPACT Institute/Food and Mood Centre, School of Medicine, Geelong, Australia
ePrimary Health Care Unit, Kuopio University Hospital, Kuopio, Finland
fSchool of Pharmacy, University of Eastern Finland, Kuopio, Finland
gInstitute of Clinical Medicine, University of Eastern Finland, Kuopio, Finland
hClinical Psychology and Psychophysiology lab, University of Cyprus, Nicosia, Cyprus
iInstitute of Clinical Medicine, University of Oslo, Oslo, Norway
jR&D department, Division of Mental Health Services, Akershus University Hospital, Lørenskog, Norway
kDepartment of Psychiatry, University of Helsinki, Helsinki, Finland
A R T I C L E I N F O Keywords:
Major depression Allostatic load Metabolic Population
A B S T R A C T
Objective: We compared the level of allostatic load (AL) between patients with major depressive disorder (MDD) and non-depressed controls using two definitions of AL: continuous AL scores (AL index) and clinically significant high AL (≥4). We examined whether MDD was associated with AL independent of basic socioeconomic (age, sex, cohabiting status and level of education) and lifestyle factors (smoking and alcohol use).
Methods: The MDD patient sample consisted of 177 psychiatric outpatients (mean age 33.7, SD 10.7 years), who were recruited from the Department of Psychiatry at Kuopio University Hospital, Finland, in 2016–19. The non- depressed controls (n =228, mean age 49.8, SD 10.1 years) lived in the municipality of Lapinlahti, Finland. Ten biomarkers were used to construct the two AL variables. These indicators were systolic and diastolic blood pressure, total cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, glucose, creatinine, waist circumference, body mass index (BMI) and C-reactive protein (CRP).
Results: The mean AL scores did not significantly differ between MDD patients (2.97) and non-depressed controls (3.12), thus it was not associated with MDD in univariate analysis. In multivariate models a higher AL index was associated with a 1.42 to 1.82 times higher likelihood of belonging to the MDD group. Furthermore, we found that high AL (i.e. AL ≥4) was associated with MDD, with the likelihood ranging between 2.27 and 2.96 compared with the non-depressed controls in multivariate models.
Conclusions: Even young adult patients with MDD appear to display clinically significant, high AL compared with non-depressed controls. Thus, it is important to pay attention to the somatic health of depressed patients in addition to their mental health.
1. Introduction
Allostatic load (AL) refers to the activation of physiological regula- tory systems in response to stress and the immediate and long-term ef- fects of these systems on the body. A stressful life results in repeated
cycles of cumulative multisystem dysregulation, which is reflected in AL. As a response to psychological and physiological stressors, the hypothalamic–pituitary–adrenal (HPA) axis is activated, resulting in the secretion of corticotropin-releasing hormone (CRH) from the hypo- thalamus [25]. AL leads to secondary systematic dysregulation of
* Corresponding author at: School of Educational Sciences and Psychology, University of Eastern Finland, P.O. BOX 111, FI-80100 Joensuu, Finland.
E-mail address: kirsi.honkalampi@uef.fi (K. Honkalampi).
Contents lists available at ScienceDirect
Journal of Psychosomatic Research
journal homepage: www.elsevier.com/locate/jpsychores
https://doi.org/10.1016/j.jpsychores.2021.110389
Received 10 June 2020; Received in revised form 27 January 2021; Accepted 1 February 2021
Journal of Psychosomatic Research 143 (2021) 110389
2 metabolic, neuroendocrinological, cardiovascular and inflammatory biomarkers [3]. These biomarkers attempt to compensate for dysregu- lated stress hormones [27]. Several studies have found that high AL can lead to various health problems and even increase the risk of mortality among older adults [10,24,28,35,36]. In contrast, however, high AL was not associated with an increased risk of all-cause mortality in a five-year follow-up among a large sample of the Scottish population with a mean age of 51 (SD 18) years [40].
Similarly to stress, depression leads to systemic physiological dys- regulation [35]. This is not surprising, since clinical depression and a prolonged stress response lead to similar changes in cognition and af- fective responses, and both states display dysregulation of autonomic functions [8]. Depression is considered to increase the risk for the pro- gression of serious somatic diseases, including cardiovascular disease, stroke, cancer and diabetes [47]. In a large sample of patients who were treated in Veteran’s Health Administration settings, depression was associated with a higher risk of three-year mortality from, for example, heart disease, respiratory illness, cerebrovascular disease and diabetes [52]. All these diseases are also connected with high AL [5,10,49]. In addition, a meta-analysis by Shi et al. (Shi et al. [45]) summarized that depression is an independent risk factor for cardiac events and mortality in individuals with and without cardiovascular disease.
In addition, Shi et al. [45] found that the somatic consequences of depression were linked to dysregulation in a range of different biological mechanisms, including autonomic, neuroendocrinological, immuno- inflammatory and metabolic mechanisms, as well as the hypothal- amic–pituitary–adrenal (HPA) axis. This is not surprising, because many similar health behaviours are associated with depression [33], as well as the development of high AL [22]. For example, a study from the Hawaii Personality and Health Cohort found that health behaviour and dysre- gulation were associated with self-rated health for both sexes [22].
The effect of the accumulation of AL varies in relation to age and many sociodemographic factors. A large population-based study, the National Health and Nutrition Examination Survey, found that AL was remarkably constant in older age groups, while it increased sharply from 20 to 60 years of age Crimmins et al. [14]. Furthermore, there could be differences between ages and sexes in underlying disease mechanisms that lead to the accumulation of AL [51]. A recent meta-analysis [20]
summarized that, in addition to this, several socioeconomic factors (low socioeconomic status, low educational level) [21] and poor lifestyle habits (smoking, alcohol consumption, poor diet) [6,32,46] have been found to be related to high AL [9].
There is some evidence for an association between depressive symptoms and depressive disorders and AL [26,30,31]. In the Douglas Hospital Longitudinal Study of Normal and Pathological Aging, an in- crease in AL was associated with a simultaneous increase in depressive symptoms in the same year of assessment among elderly people with geriatric depression. After three years of follow-up, AL was associated with depressive symptoms, but age was the only significant predictor of depressive symptoms after six years of follow-up [26]. In the Mindful- ness to Improve Elders’ Immune and Health Status (MIEIHS) study, physiological dysfunction was associated with elevated scores for af- fective, somatic and overall depressive symptoms among elderly adults Kobrosly et al. [30,31]. Berger et al. [4] investigated the relationship between symptoms of depression and two biological pathways thought to mediate depression risk – the stress hormone cortisol and allostatic load (AL) – in an Australian Aboriginal population and found that AL was selectively associated with anhedonia but not with depression. A population-based study, Health 2000 [53], demonstrated that higher burnout and cynicism and decreased professional efficacy are related to higher AL, independent of age, sex, education, occupation and psycho- logical distress. However, depression explained 60% of this association.
Thus, previous research regarding the association between AL and depression has often focused on older adults, but there is a lack of knowledge about the association between AL and depression among younger adults. Furthermore, a need for AL studies among clinical
populations has been brought up previously [7]. Studies on whether the accumulation of AL differs between depressed vs. non-depressed in- dividuals are scarce.
The aims of our study were 1) to compare AL between adult patients with major depressive disorder (MDD) and non-depressed adult controls in the general population, using two different definitions (i.e., raw AL scores and clinically significant high AL (≥4), and 2) to examine whether AL is associated with MDD, taking into account basic socio- economic (age, sex, cohabiting, education) and lifestyle factors (smok- ing, alcohol use).
We hypothesized that, similarly to an earlier study Kobrosly et al.
[30,31], patients with MDD more frequently have high AL (i.e. AL ≥4) scores and clinically significant AL compared to non-depressed controls.
However, we used socioeconomic status and lifestyle factors as control variables. Thus, we hypothesized that high AL and the AL index are associated with MDD, after taking basic socioeconomic and lifestyle factors into account.
2. Materials and methods 2.1. Participants and procedures
The study sample consisted of 177 psychiatric outpatients with MDD, whose mean (SD) age was 33.7 years (10.7, range 19–61) and who were recruited from the Department of Psychiatry at the Kuopio University Hospital, Finland, in 2016–2019. The recruitment setting was a uni- versity hospital clinic, where the participants received standard psy- chiatric outpatient care. The inclusion criterion was that the patient suffered from at least one specific mood disorder (F32.1 (n =12 in the final sample), F32.2 (n =50 in the final sample), F32.3 (n =2 in the final sample), F33.1 (n =22 in the final sample), F33.2 (n =85 in the final sample), F33.3 (n =6, in the final sample)) according to the ICD-10 diagnostic classification [39]. Patients with a diagnosis of bipolar dis- order, psychotic disorder and depression directly related to a somatic condition or substance abuse based on SCID were excluded from this study. All participants (n =186) provided written informed consent. In addition, patients with missing data (n =9; 4 men, 5 women) were excluded from the final sample, and thus it consisted of 177 MDD pa- tients. No significant differences were found in sex, mean age, smoking or alcohol use between patients with or without missing data (data not shown).
The diagnosis of MDD was confirmed by means of the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID I) [17], conducted by a trained research nurse. Those who fulfilled the study criteria and were willing to participate formed the study population. The 21-item Beck Depression Inventory (i.e., BDI) scores [1] were used to assess current depressiveness. The mean BDI score was 21.7 (SD 10.3), indi- cating a mood ranging from moderate to severe depression [2].
Of the 177 patients, 137 (77%) used antidepressive medication.
Antidepressant use was distributed as follows: 1) selective serotonin reuptake inhibitors (SSRI), n =74 (41.8%, one of the patients used two different types of SSRI antidepressants; escitalopram, n =43, 24.5%;
paroxetine, n =3, 1.7%; sertraline, n =11, 6.2%; citalopram, n =13, 7.3%; fluoxetine, n =5, 2.8%); 2) serotonin–norepinephrine reuptake inhibitors (SRNI), n =53 (29.9%, eight patients reported using two SNRI antidepressants simultaneously; venlafaxine, n =33, 18.6%; mirtaza- pine, n =20, 11.3%; duloxetine, n =8, 4.5%; and trazodone, n =2, 1.13%); 3) tricyclic antidepressants, n =5 (2.8%; doxepin, n =1, 0.6%;
amitriptyline, n =4, 2.3%); and 4) other antidepressants, n =36 (20.2%;
bupropion, n =10, 1.7%; agomelatine, n =8, 4.5%; and vortioxetine, n
=15, 8.5%). In addition, one patient (0.6%) used monoamide oxidase inhibitor (moclobemide).
The non-depressed population-based control sample consisted of 480 persons living in the municipality of Lapinlahti, which is a part of the Kuopio University Hospital catchment area in Finland. The control sample included in this study was collected as part of the five-year K. Honkalampi et al.
follow-up e Lapinlahti Study in 2010 [41]. The exclusion criteria were an elevated level of depressive symptoms, i.e., BDI scores ≥10 [2] at baseline and/or at the five-year follow-up of the Lapinlahti Study, or the reported use of antidepressants [41]. In addition, 26 participants (12 men, 14 women) were excluded because of missing data. Altogether, 228 non-depressed controls were derived from the sample. They were significantly younger than subjects with missing data (49.84, SD 10.09, range 31–66 years vs. 55.04, SD 9.28, range 31–66 years, t =2.51, df = 252, p =0.013), but there was no difference in the sex ratio between participants and subjects with vs. without missing data. The protocols for both included studies were approved by the Research Ethics Com- mittee of the Northern Savo Hospital District (Finland).
2.2. Measures
A research nurse conducted a clinical evaluation for both the MDD patients and participants from the general population. Height and body weight were measured in lightweight clothing without shoes, and body mass index (BMI; kg/m2) was calculated. Blood pressure (systolic and diastolic) and waist circumference were measured. All participants were instructed to visit the laboratory in the morning and abstain from eating during the 12 h preceding blood sampling. The levels of total choles- terol, high-density lipoprotein (HDL) cholesterol, triglycerides, glucose, creatinine and C-reactive protein (CRP) were determined using a Cobas® c501 analyser (Roche diagnostics, Penzberg, Germany) at the Eastern Finland Laboratory Centre Joint Authority Enterprise (ISLAB).
Participants completed a questionnaire about their sociodemo- graphic background: age, sex, basic education (years) and cohabiting status (unmarried, separated, divorced or widowed vs. living with a partner or married), as well as smoking (current smoking vs. no smok- ing). In addition, the participants in both samples completed a self- report questionnaire on their weekly alcohol use (less than 9 standard drinks vs. 9 or more standard drinks; 1 standard drink in Finland con- tains 12 g of ethanol).
2.3. Allostatic load
The following ten biomarkers were used to determine AL: systolic and diastolic blood pressure (i.e., indicators of cardiovascular functioning);
total cholesterol, HDL cholesterol, triglycerides, glucose and creatinine (i.e., indicators of metabolic functioning); waist circumference and BMI (i.
e., indicators of anthropometric functioning); and CRP (i.e., an indicator of immune system functioning) (Table 1).
The number of biomarkers per AL indicator varied between 1 and 5, with at least one biomarker from each category [16]. The AL was calculated as an index of physiological dysregulation for each partici- pant, with each biomarker categorized as high risk (+1), normal range or low risk (0). For those markers where the clinical cut-offs differed between males and females, sex-specific cut-offs were calculated [7].
The utilized clinically relevant cut-off points [7] are presented in Table 1. Each biomarker was dichotomized as 0 or 1, depending on its clinically relevant cut-off, and each biomarker was thus allotted an equal weight in the index. To create the AL index, in line with earlier studies [4,11], the sum of biomarkers in each category (cardiovascular, metabolic, immune, anthropometric) was divided by the number of biomarkers in each category to allow for equal weighting of the four categories. According to Juster, McEwen, & Lupien [27], this traditional count-based method for computing the AI index is utilized most frequently.
Similarly to earlier studies, the above-described AL index was used both as a continuous variable (i.e., AL scores) [4,19,26] and as a cate- gorical variable [5,18]. When used as a categorical variable, AL (i.e., (i.
e. AL ≥4) was classified into two groups, in which the cut-off score of 4 was used to denote high AL, because previous literature has demon- strated that differences in morbidity and mortality arise between groups when AL scores reach 3 or 4 [18].
2.4. Data analysis
Descriptive differences in AL (score and cut-off score) between the study sample and the control group were examined by using the χ2 test for categorical variables and t-test for continuous variables. The normal distribution of residuals (standardized and unstandardized) was verified by analysis of variance with the one-sample Kolmogorov–Smirnov test.
Spearman’s non-parametric correlation coefficients were used to calculate bivariate correlations.
Multivariable binary logistic regression analysis (method: enter) was carried out to examine whether the AL score (≥4) or high AL is asso- ciated with MDD. We constructed five models: Model 1 was a crude model only including AL. Model 2 was adjusted for age and sex, Model 3 was adjusted for age, sex, socioeconomic variables, education and cohabiting status and Model 4 was adjusted for age, sex, smoking and alcohol use. In Model 5, all the previously listed variables were included in the model.
The magnitude of the difference in the mean scores was calculated using Cohen’s d as a measure of effect size (ES) (small if d ≤0.3, medium if d =0.31 to 0.5, large if d =0.51 to 0.8, and very large if d =0.81 to 1.20) [12]. A p-value of <0.05 was considered statistically significant in all analyses. All statistical tests were two-tailed, and the analyses were conducted with the SPSS statistical package (IBM® SPSS® Statistics version 25.0).
3. Results
MDD patients were significantly younger (33.65, SD 10.74 vs. 49.84, SD 10.1 years, t =15.57, df =403, p =0.001), more often females (77.4% vs. 50.9%, χ2 =29.9, p =0.001), had a higher level of education (52.5% vs. 23.4%, χ2 =31.31, p =0.001), were less often cohabiting or married (37.3% vs. 84.6%, χ2 =96.95, p =0.001) and more frequently current smokers (48.6% vs. 17.1%, χ2 =48.14, p =0.001) than non- depressed controls. Compared to non-depressed men, male MDD pa- tients were significantly younger, more frequently current smokers and they had a higher level of education. Similar differences were found between non-depressed women and females in the MDD sample (Table 2). There was no significant difference in AL between users and Table 1
Biomarkers of allostatic load and their cut-offs based on [7].
Biomarkers Biomarker type Normal
range No risk (or lower limit)
Risk (or highest limit) Systolic blood
pressure (mmHg) Cardiovascular 90–140 ≤102.5 (0)
≥127 (1) Diastolic blood
pressure (mmHg) Cardiovascular 60–90 ≤67.5 (0)
≥82.5 (1) Waist circumference
(cm)* Anthropometric
Men Women <102
<88 NA ≥102 (1)
≥88 (1) Body Mass Index, kg/
m2 Anthropometric 18.5–25 ≤20.13
(0)
≥23.38 Total cholesterol (1)
(mmol/L) Metabolic 2.8–5.2 ≤3.4 (0) ≥4.6 (1)
HDL cholesterol
(mmol/L) Metabolic 0.9–2.0 ≤1.18
(1) ≥1.73 (0) Triglycerides mmol/L Metabolic 0.4–1.7 ≤0.75
(0)
≥1.45 (1) Glucose mmol/L Metabolic 3.9–5.5 3.9 (0) ≥5.6 (1) Creatinine (μmol/L) * Metabolic
Men Women 60–100
50–90 NA >101 (1)
>90 (1) High sensitivity C-
reactive protein (hs- CRP) (mg/L)
Inflammatory 0–8 ≤2 (0) ≥6 (1)
Allostatic load index NA 1 9
Abbreviations: CRP =C-reactive protein; HDL =High-density lipoprotein; NA = Not applicable.
Journal of Psychosomatic Research 143 (2021) 110389
4 non-users of any antidepressant medication within the MDD group (non- users of antidepressant medication: AL index 2.86, SD 1.02; MDD pa- tients using SSRIs: AL index 2.82, SD 0.89, ns; MDD patients using SRNIs: AL index 3.25, SD 0.90, ns; MDD patients using other antide- pressants: AL index 3.11, SD 1.02, ns, and using tricyclene: AL index 3.00, SD 0.96, ns).
There were no significant differences between samples in the AL mean score (Table 3, Fig. 1). The average AL score was significantly higher in men than in women among the non-depressed controls (t =
− 5.34 (226), p =0.001, ES =1.55), but no sex difference was found among MDD patients. Altogether, 14.1% (n =25) of the MDD patients and 11.4% (n =26) of the non-depressed controls belonged to the high AL group (AL ≥4).
Regarding the biomarkers of AL, MDD patients had significantly
lower systolic blood pressure, glucose levels and CRP levels than the non-depressed controls. In addition, male MDD patients had signifi- cantly higher creatinine levels and a smaller waist circumference than non-depressed males (Table 3). Male MDD patients had higher levels of HDL cholesterol and creatinine, lower levels of glucose and a smaller waist circumference compared to non-depressed males. Females in the MDD group had higher glucose, triglyceride, and hs-CRP levels than non-depressed control females (Table 3).
Age and AL score correlated significantly in both samples; the cor- relation was r =0.26 (p =0.001) in MDD patients and r =0.20 in the non-depressed controls (p =0.001).
In the logistic regression analysis (Table 4), neither the AL index nor high AL (i.e. AL ≥4) were significantly associated with the likelihood of belonging to the MDD group in the crude model. When age and sex were Table 2
Comparison of background variables between men and women in MDD and general population control samples.
Men (MDD n =
40) Men (controls) (n =
112) t (df =150), p-
value Women (MDD n =
137) Women (controls n =
116) t (df =251) and p-
value
Age (years), mean (SD) 37.13 (11.18) 49.48 (10.24) 6.39, p =0.001a 32.64 (10.42) 50.18 (9.98) 3.60, p =0.0011b
Cohabiting, n (%) 19 (47.5) 88 (78.6) χ2 =13.65, p =
0.001 47 (34.3) 105 (90.5) χ2 =82.752, p =
0.001 Education: low level, n (%) 20 (50) 96 (85.7) χ2 =20.8, p =
0.001 64 (46.7) 74 (63.8) χ2 =7.392 p =0.007
Alcohol: more than 9 doses a week
n (%) 10 (25) 16 (14.3) χ2 =2.39, p =0.12 8 (5.8) 4 (3.4) χ2 =0.802, p =0.37
Current smoker, n, (%) 25 (62.5) 22 (19.6) χ2 =5.34, p =
0.001 61 (44.5) 17 14.7) χ2 =27.582, p =
0.001 Effect size (Cohen d) a =1.15, b =1.72.
1Significance of the difference between groups was assessed with the independent samples t-test.
2Significance of the difference between groups was assessed with the Pearson chi-squared test vs. others.
Table 3
Comparison of the levels of biomarkers between samples and between men and women in the MDD and general population control samples. We have calculated effect sizes (Cohen’s d) for all significant differences between groups.
Men (MDD n
=40) Men (controls) (n
=112) t (df =150), p-
value Women (MDD
=137) Women (controls
n =116) t (df =251) p-
value MDD vs. controls (total sample) t (df =403) p-value Cardiovascular
biomarkers Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Systolic blood pressure
(mmHg) 132.48
(11.69) 148.74 (18.15) 5.28 p =
0.001a 120.93 (13.02) 139.00 (18.63) 9.04 p =
0.001b 12.00, p =0.001c Diastolic blood pressure
(mmHg) 83.83 (9.0) 86.36 (10.23) 1.39, p =0.17 81.85 (9.69) 80.95 (8.61) −0.78, p =
0.44 1.35, p =0.18
Metabolic biomarkers Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Total cholesterol (mg/dl) 5.15 (1.19) 4.95 (1.0) t-0.99, p =
0.32 5.12 (4.21) 4.96 (0.91) −0.39, p =
0.70
−0.65, p =0.52 HDL cholesterol (mg/dl) 1.42 (0.38) 1.28 (0.35) −2.04 p =
0.043d 1.66 (0.45) 1.59 (0.45) −1.55, p =
0.12 −4.07, p =0.001e Triglycerides 1.47 (1.16) 1.48 (0.93) 0.06 p =0.95 1.22 (0.75) 1.07 (0.45) −1.84, p =
0.07f 0.03, p =0.97
Glucose 5.92 (2.44) 7.04 (2.46) 2.49 p =
0.014g 5.72 (2.28) 7.19 (1.95) 5.45, p =
0.001h 5.98, p =0.001i Creatinine 80.85 (11.56) 75.5 (11.69) −2.50 p =
0.014j 63.77 (10.04) 63.63 (10.40) −0.11, p =
0.91 NA
Immune biomarkers Mean (SD) Mean (SD) Mean (SD) Mean (SD)
C-reactive protein 1.23 (1.81) 1.27 (1.89) −0.04, p =
0.97 2.80 (3.48) 1.45 (1.78) −3.77, p =
0.001k
−4.28, p =0.001l Anthropometric
biomarkers Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Waist circumference
(cm) 92.98 (9.73) 99.25 (13.51) 2.70, p =
0.008m 89.04 (17.57) 86.38 (12.45) −1.37, p =
0.17 NA
BMI 26.45 (3.51) 28.26 (5.83) 2.03, p =
0.044n 28.14 (7.71) 26.86 (5.30) −1.51, p = 0.133
−0.34, p =0.73 Allostatic load (AL index) 3.14 (0.76) 3.39 (0.69) 1.94, p =
0.054 2.88 (0.99) 2.87 (0.79) −0.13, p =
0.90 1.61, p =0.11 Effect size (Cohen d) a =1.07, b =1.12, c =1.23, d =0.38, e =0.39,f =0.24, g =0.45, h =0.69, i =0.60, j =0.46, k=0.48, l =0.41, m =0.53,n =0.37.
(1)Significance of the difference between groups was assessed with the Pearson chi-squared test vs. others; (2) Significance of the difference between groups was assessed with the independent samples t-test.
Abbreviations: BMI =body mass index.
K. Honkalampi et al.
adjusted for, the associations became significant for both AL descriptors.
In Models 3 to 5, the following were significantly associated with the likelihood of belonging to the MDD group: high AL and the AL index, a younger age, female sex, a higher level of education and living alone in Model 3; high AL and the AL index, a younger age, female sex and being a current smoker in Model 4; and high AL and the AL index, a younger age, female sex, a higher level of education, living alone and current smoking in Model 5. Thus, the results from Models 2 to 5 indicate that a higher AL index was associated with a 1.42 to 1.82 times higher likeli- hood of belonging to the MDD group. Furthermore, we found that high AL (i.e. AL ≥4) was associated with MDD, with the likelihood ranging between 2.27 and 2.96 compared with the non-depressed controls (Table 4).
4. Discussion
We observed that both the AL index and high AL (i.e. AL ≥4) were associated with a greater likelihood of belonging to the MDD group in the multivariable adjusted models. Previous evidence for an association between depressive symptoms and AL [26,30,31,34] has been obtained from elderly populations. A large population study, the National Health and Nutrition Examination Survey, found that AL was remarkably con- stant in older age groups, while it increased sharply from 20 to 60 years of age [14]. However, in our study, the MDD patients were mainly in their thirties, thus being younger compared to the populations of pre- vious studies [26,30,31]. Our results demonstrate that the accumulation of AL or its markers, especially higher levels of BMI, waist circumference and glucose levels, as more common among MDD patients, who were on average in their thirties, than among the older general population controls.
We found no difference in the levels of AL between MDD patients taking antidepressants (SSRI, SRNI, tricyclene or other antidepressants) and MDD patients not taking any antidepressants. An earlier follow-up study found that antidepressant medication worsened the metabolic syndrome of MDD patients, especially if they had metabolic syndrome at baseline [13]. Therefore, it may be of clinical relevance to pay attention to the early accumulation of AL in patients with MDD, as high AL is associated with severe adverse somatic health outcomes in the long term
[7]. In meta-analysis, Penninx et al. [38] suggested that there is systemic inflammation and hyperactivity of the HPA axis among depressed pa- tients. In addition, another systematic review summarized that depres- sion is associated with a flatter or blunted diurnal cortisol curve, which is relatively preserved across the life span [25]. Scott et al. [43] observed that depression was related to various subsequent chronic physical health conditions in a large international dataset from 17 countries. In this study, we found that although MDD patients were much younger than controls, they had certain adverse health changes that could lead to the accumulation of AL.
The mean level of AL was higher in men than in women among the non-depressed controls, but not in MDD patients. Similarly, earlier follow-up studies among elderly cohorts [44] have reported higher AL among men than women. Interestingly, a recent study among a large general population sample of African Americans found that depressive symptomatology was associated with higher AL and with its metabolic, cardiovascular, and immune components among females but not among males [19]. So far, there has been some evidence that the underlying disease mechanisms that lead to the development of high AL could differ between ethnic groups and sexes [5,44]. In our study, all the participants were Caucasian and our results cannot therefore be generalized to other ethnic groups.
We have previously shown that 88% to 98% of the MDD patients in our sample reported early experiences of emotional abuse or neglect [23]. Scheuer et al. [42] investigated whether AL mediates the associ- ation between childhood trauma and adult depression and found that the influence of physical abuse during childhood on depression in adulthood is indeed mediated by AL. This effect was moderated by age.
Another study on American Indian adults examined the association be- tween early life trauma, post-traumatic stress disorder (PTSD) and 13 physiological biomarkers [48]. The study found that early life trauma was related to PTSD, which in turn was associated with elevated AL in adulthood. Because almost all MDD patients in our sample reported early adverse experiences, we were unable to compare the levels of the AL index between those individuals who had early adversities and those who did not.
In the present study, living alone and having a higher number of years of education were more prevalent among the MDD group, which Fig. 1.Comparison of allostatic load (continuous score) between samples and sexes with 95% CI.
Journal of Psychosomatic Research 143 (2021) 110389
6 was probably because the MDD patients were from a younger genera- tion. A study among a large sample of depressed patients found that living alone, being unemployed and having a lower educational level associated independently with lowered health-related quality of life [15]. Interestingly, a meta-analysis by Wiley et al. [50] revealed some evidence of a relationship between psychosocial resources (coping, mastery, social support, relationships) and AL, but the significance of the relationships was generally small in magnitude. In addition, in the present study, non-depressed controls had higher systolic blood pressure and glucose levels and MDD patients had lower glucose levels and higher hs-CRP levels, which probably to some extent resulted from differences in age and sex between the two samples.
This study has some limitations. Firstly, the study setting was cross- sectional and thus does not make it possible to infer causal relationships with the development of AL. Secondly, the number of males in the MDD sample was relatively low and the non-depressed controls were signifi- cantly older than the MDD patients. However, all multivariable analyses were adjusted for age and sex. Thirdly, we could not calculate the waist–hip ratio or assess any neuroendocrinological biomarkers due to the lack of data on these measures. The strength of the study is that we examined AL both as a continuous variable and with clinically signifi- cant cut-offs [5,18], and thus the obtained results are comparable with several other studies. Nevertheless, as the continuous AL index allows utilising all data points, and therefore leads to greater statistical power,
it could be considered to be preferable approach in most settings; the binary AL cut-offs may, however, prove to be clinically meaningful in the future, and are therefore worthwhile to investigate. Furthermore, this study utilized a well-establish instrument (i.e., the SCID interview) to diagnose depression in the MDD sample. In addition, in the non- depressed general population sample, we carefully excluded all those subjects who had reported the use of antidepressants or an elevated level of depressive symptoms at baseline and/or at the five-year follow-up.
In summary, in this study, we found that although the enrolled MDD patients were younger than the general population controls, they already displayed adverse health changes linked to the accumulation of AL. Because of the high prevalence of depression, effective prevention strategies could have a major public health impact, not only in reducing depression, but also the depression-related burden of somatic diseases [29,37].
Declaration of interest
The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the article.
References
[1] A.T. Beck, C.H. Ward, M. Mendelson, J. Mock, J. Erbaugh, An inventory for measuring depression, Arch. Gen. Psychiatry 4 (6) (1961) 561–571, https://doi.
org/10.1001/archpsyc.1961.01710120031004.
[2] A.T. Beck, R.A. Steer, M.G. Carbin, Psychometric properties of the beck depression inventory: twenty-five years of evaluation, Clin. Psychol. Rev. 8 (1) (1988) 77–100, https://doi.org/10.1016/0272-7358(88)90050-5.
[3] T.M. Beckie, A systematic review of allostatic load, health, and health disparities, Biol. Res. Nurs. 14 (4) (2012) 311–346, https://doi.org/10.1177/
1099800412455688.
[4] M. Berger, S. Taylor, L. Harriss, S. Campbell, F. Thompson, S. Jones, R. McDermott, Hair cortisol, allostatic load, and depressive symptoms in australian aboriginal and torres strait islander people, Stress (Amsterdam, Netherlands) 22 (3) (2019) 312–320, https://doi.org/10.1080/10253890.2019.1572745.
[5] G.S. Bey, B.M. Jesdale, C.M. Ulbricht, E.O. Mick, S.D. Person, Allostatic load biomarker associations with depressive symptoms vary among US black and white women and men, Healthcare (Basel, Switzerland) 6 (3) (2018) 105, https://doi.
org/10.3390/healthcare6030105.
[6] H.A. Beydoun, S. Huang, M.A. Beydoun, S. Hossain, A.B. Zonderman, Mediating- moderating effect of allostatic load on the association between dietary approaches to stop hypertension diet and all-cause and cause-specific mortality: 2001–2010 national health and nutrition examination surveys, Nutrients 11 (10) (2019) 2311, https://doi.org/10.3390/nu11102311.
[7] G. Bizik, M. Picard, R. Nijjar, V. Tourjman, B.S. McEwen, S.J. Lupien, R. Juster, Allostatic load as a tool for monitoring physiological dysregulations and comorbidities in patients with severe mental illnesses, Harvard Rev. Psychiatry 21 (6) (2013) 296–313, https://doi.org/10.1097/HRP.0000000000000012.
[8] P. Bongiomo, Stress, fear, trauma, and distress: Underlying factors in depression, in: J.M. Greenblatt, K. Brogan (Eds.), Integrative Therapies for Depression.
Redefining Models for Assessment, Treatment and Prevention, CRC Press, 2015, pp. 67–78, https://doi.org/10.1201/b19089-10. Retrieved from.
[9] J.T. Carbone, Allostatic load and mental health: a latent class analysis of physiological dysregulation, Stress (Amsterdam, Netherlands) (2020) 1–10, https://doi.org/10.1080/10253890.2020.1813711, ahead-of-print(ahead-of- print).
[10] R. Castagn´e, V. Gar`es, M. Karimi, M. Chadeau-Hyam, P. Vineis, C. Delpierre, M. Kelly-Irving, Allostatic load and subsequent all-cause mortality: which biological markers drive the relationship? Findings from a UK birth cohort, Eur. J.
Epidemiol. 33 (5) (2018) 441–458, https://doi.org/10.1007/s10654-018-0364-1.
[11] E. Chen, G.E. Miller, M.E. Lachman, T.L. Gruenewald, T.E. Seeman, Protective factors for adults from low-childhood socioeconomic circumstances: the benefits of shift-and-persist for allostatic load, Psychosom. Med. 74 (2) (2012) 178–186, https://doi.org/10.1097/PSY.0b013e31824206fd.
[12] J. Cohen, Statistical power analysis for the behavioral sciences, 2. ed., Erlbaum, Hillsdale, NJ [u.a.], 1988.
[13] E. Corruble, K. El Asmar, S. Trabado, C. Verstuyft, B. Falissard, R. Colle, L. Becquemont, Treating major depressive episodes with antidepressants can induce or worsen metabolic syndrome: Results of the METADAP cohort, World Psychiatry 14 (3) (2015) 366–367, https://doi.org/10.1002/wps.20260.
[14] E.M. Crimmins, M. Johnston, M. Hayward, T. Seeman, Age differences in allostatic load: an index of physiological dysregulation, Exp. Gerontol. 38 (7) (2003) 731–734, https://doi.org/10.1016/S0531-5565(03)00099-8.
[15] E.J. Daly, M.H. Trivedi, S.R. Wisniewski, A.A. Nierenberg, B.N. Gaynes, D. Warden, A.J. Rush, Health-related quality of life in depression: A STARD report, Ann. Clin.
Psychiatry 22 (1) (2010) 43. Retrieved from, https://www.ncbi.nlm.nih.gov/pub med/20196982.
Table 4
Odds ratios (OR) with 95% confidence intervals (CI) of the AL index (AL continuous scores) and a categorical high AL (≥4) measure for belonging to the MDD group in logistic regression analyses.
AL index1 High AL (≥4)2
Models Variables OR (95% CIa) OR (95% CIa)
1 AL 0.87 (0.67–1.04) 1.3 (0.71–2.3)
2 AL
Age Sex (female)
1.42 (1.05–1.93)**
0.87 (0.84–0.89)
*** 3.63 (2.03–6.48)**
2.27 (1.05–4.92)*
0.87 (0.85–0.90)
*** 2.75 (1.61–4.72) 3 AL ***
Age Sex (female) Education (less than 12 years)
Living alone
1.72 (1.20–2.48)**
0.87 (0.84–0.90)
*** 3.39 (1.78–6.43)**
0.49 (0.27–0.90)*
8.53 (4.56–15.94)
**
2.56 (1.08–6.05)*
0.88 (0.86–0.91)
*** 2.70 (4.76–16.59)
** 0.49 (0.27–0.9)*
8.9 (4.76–16.59)**
4 AL
Age Sex (female) Alcohol use (>9 dose/
week) Current smoking
1.82 (1.26–2.63)**
0.86 (0.84–0.89)
*** 3.87 (2.06–7.26)**
0.52 (0.19–1.49) 6.68 (3.47–12.88)
**
2.55 (1.09–5.97)*
0.87 (0.85–0.90)
*** 3.03(1.66–5.52)**
0.55 (0.20–1.53) 6.72 (3.50–12.90) 5 AL **
Age Sex (female) Education (less than 12 years)
Living alone Alcohol use (>9 dose/
week) Current smoking
1.76 (1.20–2.58)**
0.87 (0.84–0.90)
*** 3.19 (1.62–6.27)**
0.40 (0.21–0.77)*
6.73 (3.46–13.09)
** 0.33 (0.11–1.05) 5.29 (2.57–10.88)
***
2.96 (1.20–7.33)*
0.88 (0.85–0.91)
*** 2.55 (1.32–4.9)**
0.39 (0.25–0.75)*
7.13 (3.67–13.83)
** 0.33 (0.10–1.1) 5.52 (2.68–11.36)
***
Model 1: AL; Model 2: AL, age and sex; Model 3: AL, age, sex, educational level and relationship status; Model 4: AL, age, sex, alcohol use, current smoking;
Model 5: AL, age, sex, living alone, lower educational level, alcohol use, current smoking.
95% Confidence Intervalsa; Nagelkerke R Square1 in Model 1: 0.009; in Model 2:
0.53, in Model 3: 0.63, in Model 4: 0.44, in Model 5: 0.66.
Nagelkerke R Square2 in Model 1: 0.002, in Model 2: 0.51; in Model 3: 0.62; in Model 4: 0.59; in Model 5: 0.67.
Abbreviations: MDD =major depressive disorder, p-value: * <0.05, ** <0.01,
*** <0.001.
K. Honkalampi et al.
[16] M.T. Duong, B.A. Bingham, P.C. Aldana, S.T. Chung, A.E. Sumner, Variation in the calculation of allostatic load score: 21 examples from NHANES, J. Racial Ethn.
Health Disparities 4 (3) (2017) 455–461, https://doi.org/10.1007/s40615-016- 0246-8.
[17] M.B. First, Diagnostic and statistical manual of mental disorders, 4. ed., American Psychiatric Assoc, Washington, DC, 1994.
[18] A.T. Geronimus, M. Hicken, D. Keene, J. Bound, “Weathering” and age patterns of allostatic load scores among blacks and whites in the United States, Am. J. Public Health 96 (5) (2006) 826–833, https://doi.org/10.2105/AJPH.2004.060749.
[19] S.L. Gillespie, C.M. Anderson, S. Zhao, Y. Tan, D. Kline, G. Brock, J.J. Joseph, Allostatic load in the association of depressive symptoms with incident coronary heart disease: The jackson heart study, Psychoneuroendocrinology 109 (2019) 104369, https://doi.org/10.1016/j.psyneuen.2019.06.020.
[20] J. Guidi, M. Lucente, N. Sonino, G. Fava, Allostatic load and its impact on health: a systematic review, Psychother. Psychosom. (2020) 1–17, https://doi.org/10.1159/
000510696.
[21] P.E. Gustafsson, U. Janlert, T. Theorell, H. Westerlund, A. Hammarstr¨om, Socioeconomic status over the life course and allostatic load in adulthood: results from the northern swedish cohort, J. Epidemiol. Commun. Health (1979) 65 (11) (2011) 986–992, https://doi.org/10.1136/jech.2010.108332.
[22] S.E. Hampson, L.R. Goldberg, T.M. Vogt, T.A. Hillier, J.P. Dubanoski, Using physiological dysregulation to assess global health status, J. Health Psychol. 14 (2) (2009) 232–241, https://doi.org/10.1177/1359105308100207.
[23] K. Honkalampi, N. Flink, S.M. Lehto, A. Ruusunen, H. Koivumaa-Honkanen, M. Valkonen-Korhonen, H. Viinam¨aki, Adverse childhood experiences and alexithymia in patients with major depressive disorder, Nordic J. Psychiatry 74 (1) (2019) 45–50, https://doi.org/10.1080/08039488.2019.1667430.
[24] A. Hwang, Li-Ning Peng, Y. Wen, Y. Tsai, L. Chang, Shu-Ti Chiou, Liang- Kung Chen, Predicting all-cause and cause-specific mortality by static and dynamic measurements of allostatic load: a 10-year population-based cohort study in Taiwan, J. Am. Med. Dir. Assoc. 15 (7) (2014) 490–496, https://doi.org/10.1016/
j.jamda.2014.02.001.
[25] J.J. Joseph, S.H. Golden, Cortisol dysregulation: the bidirectional link between stress, depression, and type 2 diabetes mellitus, Ann. N. Y. Acad. Sci. 1391 (1) (2017) 20–34, https://doi.org/10.1111/nyas.13217.
[26] R.P. Juster, M.F. Marin, S. Sindi, N.P. Nair, N. Nair, Y.K. Ng, S.J. Lupien, Allostatic load associations to acute, 3-year and 6-year prospective depressive symptoms in healthy older adults, Physiology & Behavior 104 (2) (2011) 360–364, https://
doi.org/10.1016/j.physbeh.2011.02.027.
[27] R. Juster, B.S. McEwen, S.J. Lupien, Allostatic load biomarkers of chronic stress and impact on health and cognition, Neurosci. Biobehav. Rev. 35 (1) (2010) 2–16, https://doi.org/10.1016/j.neubiorev.2009.10.002.
[28] A.S. Karlamangla, B.H. Singer, T.E. Seeman, Reduction in allostatic load in older adults is associated with lower all-cause mortality risk: MacArthur studies of successful aging, Psychosom. Med. 68 (3) (2006) 500–507, https://doi.org/
10.1097/01.psy.0000221270.93985.82.
[29] K. Kleine-Budde, R. Müller, W. Kawohl, A. Bramesfeld, J. Moock, J. R¨ossler, The cost of depression – a cost analysis from a large database, J. Affect. Disord. 147 (1) (2012) 137–143, https://doi.org/10.1016/j.jad.2012.10.024.
[30] R.W. Kobrosly, C.L. Seplaki, D.A. Cory-Slechta, J. Moynihan, E. Wijngaarden, Multisystem physiological dysfunction is associated with depressive symptoms in a population-based sample of older adults, Int. J. Geriatr. Psychiatry 28 (7) (2013) 718–727, https://doi.org/10.1002/gps.3878.
[31] R. Kobrosly, E. van Wijngaarden, C.L. Seplaki, D.A. Cory-Slechta, J. Moynihan, Depressive symptoms are associated with allostatic load among community- dwelling older adults, Physiol. Behav. 123 (2013) 223–230, https://doi.org/
10.1016/j.physbeh.2013.10.014.
[32] Y. Kusano, D.E. Crews, A. Iwamoto, Y. Sone, K. Aoyagi, T. Maeda, R. Leahy, Allostatic load differs by sex and diet, but not age in older japanese from the goto islands, Ann. Hum. Biol. 43 (1) (2015) 34–41, https://doi.org/10.3109/
03014460.2015.1013985.
[33] A. Lopresti, S.D. Hood, P.D. Drummond, A review of lifestyle factors that contribute to important pathways associated with major depression: diet, sleep and exercise, J. Affect. Disord. 148 (1) (2013) 12–27, https://doi.org/10.1016/j.
jad.2013.01.014.
[34] E.M. Maloney, R. Boneva, U.M. Nater, W.C. Reeves, Chronic fatigue syndrome and high allostatic load: results from a population-based case-control study in Georgia,
Psychosom. Med. 71 (5) (2009) 549–556, https://doi.org/10.1097/
PSY.0b013e3181a4fea8.
[35] B.S. McEwen, Mood disorders and allostatic load, Biol. Psychiatry 54 (3) (2003) 200–207, https://doi.org/10.1016/S0006-3223(03)00177-X.
[36] B.S. McEwen, P.J. Gianaros, Stress- and allostasis-induced brain plasticity, Annu.
Rev. Med. 62 (1) (2011) 431–445, https://doi.org/10.1146/annurev-med-052209- 100430.
[37] C. Otte, S.M. Gold, B.W. Penninx, C.M. Pariante, A. Etkin, M. Fava, A.F. Schatzberg, Major depressive disorder, Nat. Rev. Dis. Prime. 2 (2016) 16065, https://doi.org/
10.1038/nrdp.2016.65.
[38] B.W.J.H. Penninx, Y. Milaneschi, F. Lamers, N. Vogelzangs, Understanding the somatic consequences of depression: biological mechanisms and the role of depression symptom profile, BMC Med. 11 (1) (2013) 129, https://doi.org/
10.1186/1741-7015-11-129.
[39] G. Reed, S. Burns, C. Kogan, International Classification of Diseases, 2017.
[40] T. Robertson, G. Beveridge, C. Bromley, Allostatic load as a predictor of all-cause and cause-specific mortality in the general population: evidence from the scottish health survey, PLoS One 12 (8) (2017), e0183297, https://doi.org/10.1371/
journal.pone.0183297.
[41] J. Savolainen, H. Kautiainen, J. Miettola, L. Niskanen, P. M¨antyselk¨a, Low quality of life and depressive symptoms are connected with an unhealthy lifestyle, Scand.
J. Publ. Health 42 (2) (2014) 163–170, https://doi.org/10.1177/
1403494813504837.
[42] S. Scheuer, N. Wiggert, T.M. Brückl, Y. Awaloff, M. Uhr, S. Lucae, F.H. Wilhelm, Childhood abuse and depression in adulthood: The mediating role of allostatic load, Psychoneuroendocrinology 94 (2018) 134–142, https://doi.org/10.1016/j.
psyneuen.2018.04.020.
[43] K.M. Scott, C. Lim, A. Al-Hamzawi, J. Alonso, R. Bruffaerts, J.M. Caldas-de- Almeida, R.C. Kessler, Association of mental disorders with subsequent chronic physical conditions: World mental health surveys from 17 countries, JAMA Psychiatry 73 (2) (2016) 150–158, https://doi.org/10.1001/
jamapsychiatry.2015.2688.
[44] T.E. Seeman, B.H. Singer, C.D. Ryff, G. Dienberg Love, L. Levy-Storms, Social relationships, gender, and allostatic load across two age cohorts, Psychosom. Med.
64 (3) (2002) 395–406, https://doi.org/10.1097/00006842-200205000-00004.
[45] S. Shi, T. Liu, J. Liang, D. Hu, B. Yang, Depression and risk of sudden cardiac death and arrhythmias: a meta-analysis, Psychosom. Med. 79 (2) (2017) 153–161, https://doi.org/10.1097/PSY.0000000000000382.
[46] M. Sotos-Prieto, S.N. Bhupathiraju, L.M. Falc´on, X. Gao, K.L. Tucker, J. Mattei, A healthy lifestyle score is associated with cardiometabolic and neuroendocrine risk factors among puerto rican adults, J. Nutr. 145 (7) (2015) 1531–1540, https://
doi.org/10.3945/jn.114.206391.
[47] J.J. Strain, A. Sheloy, J.J. Strain, An updated electronic health record (EHR) for depression management, in: J.J. Strain, M. Blumenfield (Eds.), Depression as a Systemic Illness, Oxford University Press, New York, 2018, pp. 255–286.
[48] Z. Thayer, C. Barbosa-Leiker, M. McDonell, L. Nelson, D. Buchwald, S. Manson, Early life trauma, post-traumatic stress disorder, and allostatic load in a sample of american indian adults, Am. J. Hum. Biol. 29 (3) (2017), e22943, https://doi.org/
10.1002/ajhb.22943.
[49] I. van Deurzen, B. Vanhoutte, A longitudinal study of allostatic load in later life: the role of sex, birth cohorts, and risk accumulation, Res. Aging 41 (5) (2019) 419–442, https://doi.org/10.1177/0164027518813839.
[50] J.F. Wiley, B. Bei, J.E. Bower, A.L. Stanton, Relationship of psychosocial resources with allostatic load: a systematic review, Psychosom. Med. 79 (3) (2017) 283–292, https://doi.org/10.1097/PSY.0000000000000395.
[51] Y. Yang, M. Kozloski, Sex differences in age trajectories of physiological dysregulation: inflammation, metabolic syndrome, and allostatic load, J. Gerontol.
A Biol. Sci. Med. Sci. 66A (5) (2011) 493–500, https://doi.org/10.1093/gerona/
glr003.
[52] K. Zivin, M. Yosef, E.M. Miller, M. Valenstein, S. Duffy, H.C. Kales, H.M. Kim, Associations between depression and all-cause and cause-specific risk of death: A retrospective cohort study in the veterans health administration, Journal of Psychosomatic Research 78 (4) (2015) 324–331, https://doi.org/10.1016/j.
jpsychores.2015.01.014.
[53] T Hintsa, M Elovainio, M Jokela, K Ahola, M Virtanen, S Pirkola, Is there an independent association between burnout and increased allostatic load? Testing the contribution of psychological distress and depression, J Health Psychol 21 (8) (2016) 1576–1586.
