- Email: [email protected]
Self-Perceived Emotional Distress and Diabetes Risk Among Young Men
Self-Perceived Emotional Distress and Diabetes Risk Among Young Men Gilad Twig, MD, PhD,1,2,3,4 Hertzel C. Gerstein, MD,5 Eyal Fruchter, MD,3 Avi Shina, MD, MHA,2,3,6 Arnon Afek, MD, MHA,4,7 Estela Derazne, MSc,3,4 Dorit Tzur, MBA,3 Tali Cukierman-Yaffe, MD,4,8 Daniela Amital, MD, MHA,4,9 Howard Amital, MD, MHA,1,4 Amir Tirosh, MD, PhD2,8,10 Introduction: There are mixed data regarding the effect of emotional distress on diabetes risk, especially among young adults. This study assessed the effect of self-perceived emotional distress on diabetes incidence among young men. Methods: Incident diabetes during a mean follow-up of 6.3 (4.3) years was assessed among 32,586 men (mean age, 31.0 [5.6] years) of the Metabolic, Lifestyle, and Nutrition Assessment in Young Adults cohort with no history of diabetes between 1995 and 2011. Emotional distress was assessed by asking participants as part of a computerized questionnaire: Are you preoccupied by worries or concerns that affect your overall wellbeing? Time-dependent Cox models were applied. Data analysis took place between 2014 and 2015. Results: There were 723 cases of diabetes during 206,382 person-years. The presence of distress was associated with a 53% higher incidence of diabetes (95% CI¼1.08, 2.18, p¼0.017) after adjustment for age, BMI, fasting plasma glucose, family history of diabetes, triglyceride and high-density lipoprotein cholesterol levels, education, cognitive performance, white blood cell count, physical activity, and sleep quality. These results persisted when distress, BMI, physical activity, and smoking status were treated as time-dependent variables (hazard ratio¼1.66, 95% CI¼1.21, 2.17, p¼0.002). An adjusted hazard ratio of 2.14 (95% CI¼1.04, 4.47, p¼0.041) for incident diabetes was observed among participants persistently reporting emotional distress compared with those persistently denying it. Conclusions: Sustained emotional distress contributes to the development of diabetes among young and apparently healthy men in a time-dependent manner. These findings warrant awareness by primary caregivers when stratifying diabetes risk. (Am J Prev Med 2015;](]):]]]–]]]) & 2015 American Journal of Preventive Medicine
Introduction From the 1Department of Medicine B, Sheba Medical Center, Tel Hashomer, Israel; 2The Dr. Pinchas Bornstein Talpiot Medical Leadership Program, Sheba Medical Center, Tel Hashomer, Israel; 3The Israel Defense Forces Medical Corps, Israel; 4The Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel; 5Division of Endocrinology and Metabolism, and Population Health Research Institute, McMaster University and Hamilton Health Sciences, Hamilton, Ontario, Canada; 6Department of Obstetrics and Gynecology, Sheba Medical Center, Tel Hashomer, Israel; 7 Israel Ministry of Health, Jerusalem, Israel; 8Department of Endocrinology, Sheba Medical Center, Tel Hashomer, Israel; 9Department of Psychiatry B, Ness Ziona Mental Health Center, Ness Ziona, Israel; and 10 Division of Endocrinology, Diabetes and Hypertension, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts Address correspondence to: Gilad Twig, MD, PhD, Department of Medicine B, Chaim Sheba Medical Center, Tel Hashomer, Ramat Gan 52621, Israel. E-mail: [email protected]. 0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2015.12.006
& 2015 American Journal of Preventive Medicine
T
he incidence of Type 2 diabetes is increasing worldwide, involving an increasing number of young adults. The observation that established risk factors such as obesity cannot solely account for this increase1,2 highlights the need to identify novel risk factors. Psychosocial factors, including emotional distress, have been suggested as potential risk factors for future diabetes.3–5 However, the data are conflicting, with some prospective studies reporting a twofold or higher risk of diabetes in people reporting excessive distress,6–8 whereas others report equivocal risk9,10 or even a risk reduction.11 Each of these studies defined emotional distress differently while utilizing various approaches to adjust for confounders, with limited consideration of distress level changes over time. Moreover, these studies
Published by Elsevier Inc.
Am J Prev Med 2015;](]):]]]–]]] 1
2
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
usually relied on a single baseline assessment in middleaged individuals in their sixth decade of life or older. The Metabolic, Lifestyle, and Nutrition Assessment in Young Adults (MELANY) cohort is an ongoing, large, prospective study in Israel assessing the effect of clinical and biochemical parameters on cardiometabolic morbidity among young adults.12–15 Data from more than 32,000 young healthy men were analyzed to explore the relationship between emotional distress and incident diabetes after accounting for a comprehensive set of confounders and after considering changes in distress level over time.
Methods The Metabolic, Lifestyle, and Nutritional Assessment in Young Adults Cohort The MELANY cohort is being conducted at the Israel Defense Forces screening center (Zrifin, Israel), to which all career service personnel aged 425 years are referred every 3–5 years for routine health examinations and screening tests.12,16 At each visit to the screening center, participants complete a detailed questionnaire in Hebrew assessing nutritional, medical, and lifestyle factors, including a brief question regarding self-perceived emotional distress. Blood samples are drawn and analyzed immediately following a 14-hour fast. Complete physical examination is performed by a physician at the center and includes measurements of blood pressure, weight, and height. Primary care for all Israel Defense Forces personnel between scheduled visits to the center is obtained at designated military clinics, and all medical information is recorded in a central computerized database.
Study Population Figure 1 depicts the study design. Between 1995 and 2011, all firsttime attendees examined at the screening center were asked the
following question: Are you preoccupied by worries or concerns that affect your overall wellbeing? This analysis included all men who answered either yes, no, or I am not sure; had a fasting plasma glucose (FPG) level o126 mg/dL; and denied a history of depression or diabetes. The IRB of the Israeli Defense Medical Corps approved this study with the assurance of strict participant anonymity during data analyses.
Measures Participants were followed prospectively from enrollment (first visit to the screening center) and the primary outcome was diabetes onset. Participants underwent the same biochemical and clinical evaluation (including screening for emotional distress and FPG-based screening for diabetes) at each consecutive visit. All subjects were censored at the time of death, retirement from military service, March 8, 2011, or diabetes diagnosis, whichever came first. Incident cases of diabetes were diagnosed by a physician according to the American Diabetes Association diagnostic criteria, by documenting either two FPG levels Z126 mg/dL or a glucose level Z200 mg/dL 2 hours after ingestion of 75 grams of glucose. No antibody data were available; as such, the type of diabetes (e.g., Type 1) could not be ascertained. However, it was recently reported for this cohort that 498% of diabetes diagnosed cases were not prescribed insulin during the first year, thereby supporting Type 2 predominance. All laboratory studies were performed on fresh samples in an ISO-9002 quality-assured core facility laboratory. Emotional distress was assessed by the question: Are you preoccupied by worries or concerns that affect your overall wellbeing? (yes, no, uncertain). The assessment of distress by similar single-item measures exhibited good test–retest reliability and good correlation with multi-item distress questionnaires,17 and was used to establish an association with different clinical outcomes,18–20 including diabetes.6 The Mini Sleep Questionnaire (MSQ) is a ten-question questionnaire used in clinical studies to assess sleep quality.21–23 Cognitive performance at pre-recruitment evaluation (approximately at age 17 years) was denoted by a
Figure 1. Diagram of study design and outcomes. www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]] 24
general intelligence score (GIS) as previously described. The GIS was shown to correlate well with the Wechsler Adult Intelligence Scale total IQ25 and was found to serve as an independent risk factor for diabetes in this cohort in earlier analyses.24,26 Both MSQ and GIS were modeled as continuous variables. SES data were coded by the area of residence and was stratified according to a 1– 10 scale for all municipalities in Israel as devised by the Israeli Central Bureau of Statistics.27,28 SES was categorized into the groups: low (1–4); medium (5–7); and high (8–10); as reported previously.27,28 Marital status (single, living with a spouse, previously married and currently living alone); consumption of sleeping or relaxation pills (no, yes); type of army service (field versus home front predominant); and education level (r8, r12, Z13 years of education) were treated as categorical variables. Country of origin (classified by the father’s or grandfather’s country of birth) was categorized into five geographic areas:13,24 former USSR countries; Asia (non-USSR); Africa (excluding South Africa); Western (comprised of non-USSR Europe, North and South America, South Africa, Australia, and New Zealand); and Israel. Birth country (indicating nativity rather than ancestry) was classified in a similar manner. Statistical analysis was conducted between 2014 and 2015. Age; triglyceride level; FPG; high-density lipoprotein cholesterol (HDL-c); and white blood cell count were treated as continuous variables. Physical activity (not active, o150 minutes/week, Z150 minutes/week); smoking status (never, ex-smoker, current); family history of diabetes (yes, no); regular consumption of alcoholic beverages (yes, no); and regular breakfast consumption (yes, no) were treated as categorical variables. BMI was treated as a continuous variable in multivariate analysis, and as a categorical variable (normal, BMI o25; overweight, 25r BMI o30; obese, BMI Z30) assessing an interaction between BMI and distress level with future diabetes risk.
Statistical Analysis Continuous variables were summarized using means and SDs or medians with intraquartile ranges. Counts with percentages were used for binary variables. Comparison of baseline characteristics was conducted using unpaired t-tests and chi-square tests for continuous and categorical variables, respectively. Cox regression models were used to estimate the hazard ratios (HRs) and 95% CIs for diabetes after adjustment for age alone (Model 1) and after adjustment for additional clusters of risk factors (Models 1A–1D), including genetic (age, country of origin, family history of diabetes); lifestyle-related (age, physical activity, smoking status, alcohol consumption, breakfast consumption); metabolic (age, BMI, FPG, HDL-c, WBC count, triglyceride level); and psychosocial (age, type of army service, SES, GIS, education, and MSQ score). Variables that were significant at po0.05 in the ageadjusted analyses were included in the final multivariable analysis. A forward likelihood ratio stepwise model was applied for the covariates listed in the final multivariate model to rank the contribution of emotional distress with respect to other modifiable risk factors. In this analysis, non-modifiable diabetes risk factors were first introduced (age, country of origin, family history of diabetes, cognitive performance) and then the contribution of the modifiable risk was determined (education, BMI, FPG, physical activity, triglyceride level, HDL-c level, MSQ score). Log minus log plots for each categorical variable were inspected to verify the ] 2015
3
assumption of proportionality of the hazards. The final multivariate model was also performed with terms of interaction between the continuous variables and log(time) (SAS, version 13.1, PROC PHREG). Distress status, BMI, smoking status, and physical activity were also treated as time-dependent variables as appropriate. Differences in follow-up according to degree of emotional distress were compared with ANOVA and Dunnet T3 post hoc multiple comparison tests. Two supplemental analyses were done. The first was limited to participants who attended the screening center at least twice, and defined diabetes based on an FPG value Z126 mg/ dL at scheduled visits only. This was done to account for the possibility of excess consumption of healthcare services by participants reporting distress. The second was limited to participants with a minimal follow-up duration of 6 years, to account for differential follow-up of participants with and without emotional distress. Participants with missing covariate data (final model, 13.7%) were excluded from multivariable analysis. A separate analysis applied imputation for missing values using a decision tree algorithm for categorical variables and distribution methods for continuous variables (SAS Miner, version 13.1). Values represent the mean (SD), unless mentioned otherwise. Analyses were performed with SPSS, version 21.0, unless stated otherwise.
Results Table 1 presents the baseline characteristics of the cohort. Emotional distress was reported by 1,341 participants (4.1%). This subpopulation was characterized by an increase in BMI (25.8 vs 25.4, p¼0.002) and higher obesity rates (15.7% vs 12.0%, po0.001); lower degree of physical activity; and higher degree of current (or past smoking) as compared with those denying distress. FPG, prevalence of impaired fasting glucose, HDL-c, and blood pressure were comparable at baseline between groups. Participants who reported emotional distress had a threefold increase in the prevalence of abnormal sleep score (MSQ score 424, 67% vs 22%). Accordingly, consumption of sleeping and relaxation pills was more than fivefold higher among those reporting emotional distress compared with those denying it (7.9% vs 1.4%, respectively). There were 723 new cases of diabetes diagnosed during 206,382 person-years of follow-up (mean total follow-up of 6.3 [4.3] years). The incident rate of diabetes among those denying any emotional distress was 3.32 cases/ 1,000 person-years, and was nearly doubled when emotional distress was reported (6.35 cases/1,000 personyears, Table 2). This difference persisted in various models adjusting for various genetic, lifestyle, and metabolic risk factors for diabetes (Appendix Figure 1, available online). There was an HR of 1.53 attributed to emotional distress (95% CI¼1.08, 2.18, p¼0.017, Table 2) in a multivariate model adjusted for age, BMI, FPG, family history of diabetes, triglyceride level, HDL-c,
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
4
Table 1. Baseline Characteristics of the Cohort at Enrollment (First Visit to the Screening Center) Are you preoccupied by worries or concerns that affect your overall wellbeing? Characteristic n Age (years)
No
Uncertain
Yes
Total
29,797
1,447
1,341
32,586
30.9⫾5.6
31.4⫾6.0
31.6⫾6.0
31.0⫾5.6
Country of origin
p-value
o0.001 0.112
Israel
8
9
8
8
Former USSR countries
12
14
17
12
Asia
24
24
23
24
Africa
31
32
31
31
West
25
21
21
25
Marital status
0.042
Single
21.2
24.8
22.5
21.4
Living with a spouse
76.7
72.9
71.7
76.3
Divorced, living alone
2.2
2.3
5.7
2.3
SES
0.091
Low
32
32
33
Medium
53
55
54
High
15
13
13
81.4
75.1
77.6
Z13 years of education
80.9
Type of service
0.006 0.070
Field-dominant
38.4
36.1
34.5
38.1
Desk-dominant
61.6
63.9
65.5
61.9
25.43⫾4.0
25.50⫾4.1
25.82⫾4.2
25.45⫾4.0
Normal weight
50
50
46.5
50
Overweight
38
36
37.8
38
Obese
12
14
15.7
12
89.5⫾9.0
89.6⫾9.1
89.4⫾9.3
89.5⫾9.0
0.742
11.9
13.1
11.5
11.9
0.345
117.9⫾12.5
117.6⫾12.4
117.8⫾12.2
117.5⫾12.5
0.171
BPDiastolic
73.6 ⫾9.6
74.6⫾ 10.2
73.9 ⫾10.0
74.1⫾9.5
0.569
HDL-c (mg/dL)
46.3⫾10.7
45.9⫾10.5
46.0⫾11.1
46.2⫾10.7
0.237
101 (71, 147)
99 (72, 152)
105 (74, 154)
101 (71, 148)
0.072
117.2⫾33.1
118.1⫾33.0
116.6⫾34.6
117.2⫾33.2
0.574
6.65⫾1.49
6.72⫾1.49
6.84⫾1.56
6.66⫾1.49
o0.001
BMI
Fasting glucose level (mg/dL) Impaired fasting glucose BPSystolic (mean ⫾ SD, mmHg)
Triglycerides (mg/dL) (25th, 75th) LDL (mg/dL) 3
3
WBC count (10 cells/mm )
0.002
o0.001
Physical activity None
64
74
78
65
o150 minutes/week
28
21
17
27
Z150 minutes/week
8
5
5
8 (continued on next page)
www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
5
Table 1. Baseline Characteristics of the Cohort at Enrollment (First Visit to the Screening Center) (continued) Are you preoccupied by worries or concerns that affect your overall wellbeing? Characteristic Family history of diabetes
No
Uncertain
Yes
Total
p-value
13.8
19.1
18.7
14.2
o0.001 o0.001
Smoking status Never
60
52
50
59
Ex-smoker
13
15
13
13
Current
27
33
37
28
Abnormal MSQ score
23
59
68
26
o0.001
Use of sleeping pills
1.4
7.9
5.9
1.9
o0.001
Breakfast consumption
80
85
83
81
0.030
Alcohol consumption
8.0
11.5
11.5
8.2
o0.001
Note: Boldface indicates statistical significance (po0.05). BP, blood pressure; HDL-c, high-density lipoprotein cholesterol; LDL, low-density lipoprotein cholesterol; MSQ, Mini Sleep Questionnaire; USSR, Union of Soviet Socialist Republic; WBC, white blood cell.
education, GIS, white blood cell count, physical activity, and MSQ score. In a forward stepwise model, emotional distress was introduced to the model in the sixth step among modifiable risk factors, preceded by FPG, BMI, MSQ score, education, and triglyceride level (change in
chi square, 7.489; p¼0.024). Figure 2A depicts the cumulative diabetes incidence in the group with reported emotional distress at baseline versus those without. Similar results were obtained when the definition of diabetes was limited to those with an FPG Z126 mg/dL
Table 2. Hazard Ratio (HR) for Developing Diabetes Depending on Baseline Distress Level Are you preoccupied by worries or concerns that affect your overall wellbeing? Variable
No
Uncertain
Yes
Total
29,797
1,447
1,341
32,586
New cases of diabetes
635
42
46
723
Mean follow-up (years)
6.41⫾4.29
5.66⫾4.18
5.40⫾4.030
6.33⫾4.28
190,943
8,190
7,239
206,382
3.32
5.12
6.35
3.50
1 (ref)
1.624
1.899
95% CI
1.188, 2.221
1.407, 2.562
p-value
0.002
o0.001
n
Cumulative follow-up (person-years) Rate (1/1,000 person-years) Model 1: Age HR
Model 2: Age, BMI, FPG, family history of diabetes, triglyceride level, HDL-c, education, GIS, WBC, physical activity, MSQ sleep quality score HR
1 (ref)
1.361
1.530
95% CI
0.789, 1.676
1.081, 2.183
p-value
0.467
0.017
FPG, fasting plasma glucose; GIS, general intelligence score; HDL-c, high-density lipoprotein cholesterol; HR, hazard ratio; MSQ, Mini Sleep Questionnaire; WBC, white blood cells.
] 2015
6
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
Figure 2. Emotional distress level over time and diabetes incidence. Note: Incidence of cumulative diabetes over time for a single evaluation (at enrollment) and for two consequent evaluations of emotional distress level (a and b respectively). Diabetes risk was adjusted for the multivariable model indicated in Table 2 (model 2). Detailed hazard ratio and follow-up data for (a) are shown in Table 2, and for (b) in Appendix Table 2 (available online).
at scheduled follow-up visits (n¼242 cases of diabetes; occurrence of emotional distress, HR¼1.85, 95% CI¼1.19, 2.87, p¼0.006), or when only participants with a follow-up of 46 years were included (339 of 12,831 participants, HR¼1.92, 95% CI¼1.22, 3.01, p¼0.005). Of note, participants reporting emotional distress had a higher risk for diabetes than those denying it despite a shorter follow-up duration (po0.001, Table 2). Similar results were obtained for Models 1A–1D and Model 2 when imputation for missing variables was applied, or when interaction between continuous variables and log (time) was introduced into the model (data not shown). As noted in Appendix Table 1 (available online), the ageadjusted diabetes incidence of 2,914 men who did not provide an answer was similar to those who did reply. Increased body weight has been suggested as a potential link between psychosocial distress and increased diabetes risk.29 However, no attenuation in the distress–diabetes association was found when BMI was treated as a time-dependent variable (HR [Model 2]¼1.70, 95% CI¼1.21, 2.38, p¼0.002). There was no interaction between BMI and level of emotional distress with future diabetes incidence (p¼0.393). Data regarding emotional distress at two time points were available for 10,709 participants of the study population, with a minimal time interval of 2 years between scheduled visits and a subsequent follow-up of 9.50 (3.3) years (median, 9.99 years). After two consecutive visits, 8,935 participants (83.4%) consistently denied emotional distress (no–no), whereas 117 participants (1.1%) consistently confirmed it (yes–yes). Of the
442 newly diagnosed cases of diabetes, 3.76 cases/1,000 person-years occurred in the no–no group, whereas 7.69 cases/1,000 person-years occurred in the yes–yes group. This difference was translated to an adjusted HR of 2.14 (Model 2, 95% CI¼1.04, 4.47, p¼0.041) in those reporting sustained emotional distress compared with those consistently denying it (Figure 2B). Diabetes risk was not statistically different among individuals reporting decrease in emotional distress level (yes–no) and those that denied distress (Model 2, HR¼0.92, 95% CI¼0.43, 1.97, p¼0.836). When the level of emotional distress was treated as a time-dependent variable, an HR of 1.70 (95% CI¼1.25, 2.31, p¼0.001) was obtained for incident diabetes, compared with cases where it was absent. The distress–diabetes association remained significant when BMI, physical activity, and smoking status were treated as time-dependent variables (Model 2, HR¼1.66, 95% CI¼1.21, 2.17, p¼0.002). Of note, the age of diabetes onset among participants with two consequent evaluations compared with those with one evaluation was mildly delayed (41.1 [3.9] years vs 37.8 [5.0] years, po0.001).
Discussion This analysis of 32,586 young men followed for more than 200,000 person-years demonstrates an association between self-perceived emotional distress and incident diabetes. Emotional distress was found in 4.1% of the cohort and increased the risk for incident diabetes by 1.53-fold, independent of other known clinical and www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
biochemical diabetes risk factors. When emotional distress persisted in two consecutive visits, cumulative diabetes incidence exceeded 15%, whereas normalization of the distress level restored the diabetes incidence to levels similar to those without distress. Accordingly, emotional distress level was also found to be a significant risk factor in time-dependent analysis. A variety of mechanisms may mediate the distress– diabetes association. Psychological distress may dysregulate the nervous system and hypothalamic–pituitary axis, resulting in increased levels of cortisol, adrenaline, and neuropeptide Y.30,31 Abnormally elevated levels of these stress hormones trigger gluconeogenesis and lipolysis while suppressing insulin secretion and decreasing insulin sensitivity.32,33 Furthermore, distress is associated with low-grade chronic inflammatory response that can further aggravate insulin resistance.30 The observational nature of this study precludes characterization of these mechanisms. The present analyses have unique strengths. As opposed to previous studies,34–36 metabolism-related risk factors including time-dependent effect of obesity status and smoking were considered,37 consistent with a recent large study.5 Also, the authors assessed the continuous effect of emotional distress in a time-dependent manner rather than a single baseline assessment.5,29,36,37 In order to address the possibility that emotional distress may increase healthcare service use, thereby increasing the likelihood of diabetes diagnosis (sampling bias), analyses restricted to diabetes diagnosis defined as FPG Z126 mg/ dL measured at scheduled visits to the screening center were performed. Additionally, this study examined the possibility that low SES38 may confound the link between emotional distress and incident diabetes. However, the distress–diabetes relationship persisted after accounting for socioeconomic-related covariates (Table 2; Appendix Figure 1, available online). Finally, although the relationship between depressive symptoms and disturbed glucose levels may be bidirectional,35 it is unlikely that higher levels of emotional distress were due to undiagnosed diabetes at baseline. This is supported by the comparable glucose level at baseline in those with and without emotional distress and the persistence of the distress–diabetes association even after a 6-year “wash-out” period.
Limitations The limitations of this study include the use of selfreported, rather than objective, data to assess for emotional distress level. Also, evaluation of distress by an implicit question, as used in the current and previous reports,6,7 may suffer from suboptimal sensitivity. Nevertheless, single-item measures of emotional distress are ] 2015
7 17
straightforward with good test–retest reliability, easy to use in various clinical settings,39–41 and correlate well with multi-item questionnaires assessing distress.17 Additionally, the association between distress and a variety of other clinical metabolic outcomes, such as ischemic heart disease has also been established using single-item measures.18–20 Second, the MELANY cohort is restricted to a unique group of army recruits. Yet, the current findings were independent of the type of military service (Appendix Figure 1, available online), and the characteristics of the participants of the MELANY cohort were highly similar to other U.S.- and European-based cohorts of young men42–44 and are considered to adequately reflect the population of young adults in Israel.28 The characteristics and design of our cohort are also advantageous in minimizing the effect of socioeconomic confounders and by including repetitive measurements of clinical and metabolic variables with no loss to follow-up.24 Third, the analysis was conducted only in men, thereby limiting the generalization of the results to women. Finally, the distress–diabetes association is complex and may be accompanied by additional confounders that were not, or only partially, adjusted for. Nevertheless, the applied multivariate model (Table 2) represents a detailed clinical and biochemical assessment that is acceptable in clinical practice.
Conclusions This study demonstrates an association between selfperceived emotional distress over time and increased risk for diabetes among apparently young healthy men. These results emphasize the need for awareness by physicians when stratifying the risk for diabetes. This study was supported by a research grant of the Israel Defense Forces Medical Corp and the Israeli ministry of defense. GT and AT were partially supported by a grant from the Pinchas Borenstein Talpiot Medical Leadership Program, Sheba Medical Center, Tel Hashomer, Israel. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions included the following: study concept and design (GT); data acquisition (DT); statistical analyses (GT, ED); interpretation of data (GT, HCG, AS, ED, HA, EF, TCY, AT); writing the first draft of the manuscript (GT); and critical revision of the manuscript (AA, AS, AT, ED, DA, TCY, HCG, EF). The content of this manuscript was presented in the 75th American Diabetes Association meeting (Boston, MA). GT had full access to all the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis. No financial disclosures were reported by the authors of this paper.
8
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
References 1. Menke A, Rust KF, Fradkin J, Cheng YJ, Cowie CC. Associations between trends in race/ethnicity, aging, and body mass index with diabetes prevalence in the United States: a series of cross-sectional studies. Ann Intern Med. 2014;161(5):328–335. http://dx.doi.org/ 10.7326/M14-0286. 2. Lyssenko V, Laakso M. Genetic screening for the risk of type 2 diabetes: worthless or valuable? Diabetes Care. 2013;36(suppl 2):S120– S126. http://dx.doi.org/10.2337/dcS13-2009. 3. Chandola T, Brunner E, Marmot M. Chronic stress at work and the metabolic syndrome: prospective study. BMJ. 2006;332(7540):521–525. http://dx.doi.org/10.1136/bmj.38693.435301.80. 4. Bergmann N, Gyntelberg F, Faber J. The appraisal of chronic stress and the development of the metabolic syndrome: a systematic review of prospective cohort studies. Endocr Connect. 2014;3(2):R55–R80. http: //dx.doi.org/10.1530/EC-14-0031. 5. Virtanen M, Ferrie JE, Tabak AG, et al. Psychological distress and incidence of type 2 diabetes in high-risk and low-risk populations: the Whitehall II Cohort Study. Diabetes Care. 2014;37(8):2091–2097. http://dx.doi.org/10.2337/dc13-2725. 6. Rod NH, Gronbaek M, Schnohr P, Prescott E, Kristensen TS. Perceived stress as a risk factor for changes in health behaviour and cardiac risk profile: a longitudinal study. J Intern Med. 2009;266(5):467–475. http://dx.doi.org/10.1111/j.1365-2796.2009.02124.x. 7. Novak M, Bjorck L, Giang KW, Heden-Stahl C, Wilhelmsen L, Rosengren A. Perceived stress and incidence of Type 2 diabetes: a 35-year follow-up study of middle-aged Swedish men. Diabet Med. 2013;30(1):e8–e16. http://dx.doi.org/10.1111/dme.12037. 8. Eriksson AK, Ekbom A, Granath F, Hilding A, Efendic S, Ostenson CG. Psychological distress and risk of pre-diabetes and Type 2 diabetes in a prospective study of Swedish middle-aged men and women. Diabet Med. 2008;25(7):834–842. http://dx.doi.org/10.1111/j.1464-5491.2008.02463.x. 9. Williams ED, Magliano DJ, Tapp RJ, Oldenburg BF, Shaw JE. Psychosocial stress predicts abnormal glucose metabolism: the Australian Diabetes, Obesity and Lifestyle (AusDiab) study. Ann Behav Med. 2013;46(1):62–72. http://dx.doi.org/10.1007/s12160-013-9473-y. 10. Mommersteeg PM, Herr R, Zijlstra WP, Schneider S, Pouwer F. Higher levels of psychological distress are associated with a higher risk of incident diabetes during 18 year follow-up: results from the British Household Panel Survey. BMC Public Health. 2012;12:1109. http://dx. doi.org/10.1186/1471-2458-12-1109. 11. Eriksson AK, van den Donk M, Hilding A, Ostenson CG. Work stress, sense of coherence, and risk of type 2 diabetes in a prospective study of middle-aged Swedish men and women. Diabetes Care. 2013;36(9): 2683–2689. http://dx.doi.org/10.2337/dc12-1738. 12. Tirosh A, Shai I, Tekes-Manova D, et al. Normal fasting plasma glucose levels and type 2 diabetes in young men. N Engl J Med. 2005;353(14): 1454–1462. http://dx.doi.org/10.1056/NEJMoa050080. 13. Twig G, Afek A, Derazne E, et al. Diabetes risk among overweight and obese metabolically healthy young adults. Diabetes Care. 2014;37(11): 2989–2995. http://dx.doi.org/10.2337/dc14-0869. 14. Twig G, Gerstein HC, Ben-Ami Shor D, et al. Coronary artery disease risk among obese metabolically healthy young men. Eur J Endocrinol. 2015;173(3):305–312. http://dx.doi.org/10.1530/EJE-15-0284. 15. Furer A, Afek A, Beer Z, et al. Height at late adolescence and incident diabetes among young men. PLoS One. 2015;10(8):e0136464. http://dx. doi.org/10.1371/journal.pone.0136464. 16. Tirosh A, Shai I, Afek A, et al. Adolescent BMI trajectory and risk of diabetes versus coronary disease. N Engl J Med. 2011;364(14):1315– 1325. http://dx.doi.org/10.1056/NEJMoa1006992. 17. Littman AJ, White E, Satia JA, Bowen DJ, Kristal AR. Reliability and validity of 2 single-item measures of psychosocial stress. Epidemiology. 2006;17(4):398–403. http://dx.doi.org/10.1097/01.ede.0000219721. 89552.51.
18. Nielsen NR, Kristensen TS, Prescott E, Larsen KS, Schnohr P, Gronbaek M. Perceived stress and risk of ischemic heart disease: causation or bias? Epidemiology. 2006;17(4):391–397. http://dx.doi. org/10.1097/01.ede.0000220556.86419.76. 19. Nielsen NR, Kristensen TS, Schnohr P, Gronbaek M. Perceived stress and cause-specific mortality among men and women: results from a prospective cohort study. Am J Epidemiol. 2008;168(5):481–491; discussion 492–486. 10.1093/aje/kwn157. 20. Non AL, Rimm EB, Kawachi I, Rewak MA, Kubzansky LD. The effects of stress at work and at home on inflammation and endothelial dysfunction. PLoS One. 2014;9(4):e94474. http://dx.doi.org/10.1371/ journal.pone.0094474. 21. Falavigna A, de Souza Bezerra ML, Teles AR, et al. Sleep disorders among undergraduate students in Southern Brazil. Sleep Breath. 2011;15(3):519–524. http://dx.doi.org/10.1007/s11325-010-0396-6. 22. Melamed S, Oksenberg A. Excessive daytime sleepiness and risk of occupational injuries in non-shift daytime workers. Sleep. 2002;25(3): 315–322. 23. Twig G, Shina A, Afek A, et al. Sleep quality and risk of diabetes and coronary artery disease among young men. Acta Diabetol. 2015. http: //dx.doi.org/10.1007/s00592-015-0779-z. 24. Twig G, Gluzman I, Tirosh A, et al. Cognitive function and the risk for diabetes among young men. Diabetes Care. 2014;37(11):2982–2988. http://dx.doi.org/10.2337/dc14-0715. 25. Reichenberg A, Weiser M, Rabinowitz J, et al. A population-based cohort study of premorbid intellectual, language, and behavioral functioning in patients with schizophrenia, schizoaffective disorder, and nonpsychotic bipolar disorder. Am J Psychiatry. 2002;159(12): 2027–2035. http://dx.doi.org/10.1176/appi.ajp.159.12.2027. 26. Cukierman-Yaffe T, Kasher-Meron M, Fruchter E, et al. Cognitive performance at late adolescence and the risk for impaired fasting glucose among young adults. J Clin Endocrinol Metab. 2015;100 (12):4409–4416. http://dx.doi.org/10.1210/jc.2015-2012. 27. Twig G, Afek A, Shamiss A, et al. Adolescence BMI and trends in adulthood mortality: a study of 2.16 million adolescents. J Clin Endocrinol Metab. 2014;99(6):2095–2103. http://dx.doi.org/10.1210/jc.2014-1213. 28. Twig G, Livneh A, Vivante A, et al. Cardiovascular and metabolic risk factors in inherited autoinflammation. J Clin Endocrinol Metab. 2014;99(10):E2123–E2128. http://dx.doi.org/10.1210/jc.2014-2096. 29. Heraclides AM, Chandola T, Witte DR, Brunner EJ. Work stress, obesity and the risk of type 2 diabetes: gender-specific bidirectional effect in the Whitehall II study. Obesity (Silver Spring). 2012;20(2):428– 433. http://dx.doi.org/10.1038/oby.2011.95. 30. McEwen BS. Protective and damaging effects of stress mediators. N Engl J Med. 1998;338(3):171–179. http://dx.doi.org/10.1056/NEJM 199801153380307. 31. Bjorntorp P. Body fat distribution, insulin resistance, and metabolic diseases. Nutrition. 1997;13(9):795–803. http://dx.doi.org/10.1016/S08999007(97)00191-3. 32. Lundberg U. Stress hormones in health and illness: the roles of work and gender. Psychoneuroendocrinology. 2005;30(10):1017–1021. http: //dx.doi.org/10.1016/j.psyneuen.2005.03.014. 33. Bhathena SJ. Relationship between fatty acids and the endocrine and neuroendocrine system. Nutr Neurosci. 2006;9(1-2):1–10. http://dx. doi.org/10.1080/10284150600627128. 34. Knol MJ, Twisk JW, Beekman AT, Heine RJ, Snoek FJ, Pouwer F. Depression as a risk factor for the onset of type 2 diabetes mellitus. A meta-analysis. Diabetologia. 2006;49(5):837–845. http://dx.doi.org/ 10.1007/s00125-006-0159-x. 35. Mezuk B, Eaton WW, Albrecht S, Golden SH. Depression and type 2 diabetes over the lifespan: a meta-analysis. Diabetes Care. 2008; 31(12):2383–2390. http://dx.doi.org/10.2337/dc08-0985. 36. Nyberg ST, Fransson EI, Heikkila K, et al. Job strain as a risk factor for type 2 diabetes: a pooled analysis of 124,808 men and women. Diabetes Care. 2014;37(8):2268–2275. http://dx.doi.org/10.2337/dc13-2936.
www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]] 37. Mooy JM, de Vries H, Grootenhuis PA, Bouter LM, Heine RJ. Major stressful life events in relation to prevalence of undetected type 2 diabetes: the Hoorn Study. Diabetes Care. 2000;23(2):197–201. http: //dx.doi.org/10.2337/diacare.23.2.197. 38. Kumari M, Head J, Marmot M. Prospective study of social and other risk factors for incidence of type 2 diabetes in the Whitehall II study. Arch Intern Med. 2004;164(17):1873–1880. http://dx.doi.org/10.1001/ archinte.164.17.1873. 39. Hermelink K, Hohn H, Hasmuller S, et al. Brief distress screening in clinical practice: does it help to effectively allocate psycho-oncological support to female cancer inpatients? Breast Care (Basel). 2014;9(2): 129–133. http://dx.doi.org/10.1159/000360788. 40. Schubart JR, Mitchell AJ, Dietrich L, Gusani NJ. Accuracy of the Emotion Thermometers (ET) screening tool in patients undergoing surgery for upper gastrointestinal malignancies. J Psychosoc Oncol. 2015;33(1):1–14. http://dx.doi.org/10.1080/07347332.2014. 977415. 41. Keir ST, Calhoun-Eagan RD, Swartz JJ, Saleh OA, Friedman HS. Screening for distress in patients with brain cancer using the NCCNʼs
] 2015
9
rapid screening measure. Psychooncology. 2008;17(6):621–625. http: //dx.doi.org/10.1002/pon.1271. 42. Vassy JL, Hivert MF, Porneala B, et al. Polygenic type 2 diabetes prediction at the limit of common variant detection. Diabetes. 2014;63 (6):2172–2182. http://dx.doi.org/10.2337/db13-1663. 43. Wurtz P, Soininen P, Kangas AJ, et al. Branched-chain and aromatic amino acids are predictors of insulin resistance in young adults. Diabetes Care. 2013;36(3):648–655. http://dx.doi.org/10.2337/dc12-0895. 44. Nguyen QM, Srinivasan SR, Xu JH, et al. Elevated liver function enzymes are related to the development of prediabetes and type 2 diabetes in younger adults: the Bogalusa Heart Study. Diabetes Care. 2011;34(12):2603–2607. http://dx.doi.org/10.2337/dc11-0919.
Appendix Supporting data Supplementary data associated with this article can be found at http://dx.doi.org/10.1016/j.amepre.2015.12.006.
Introduction From the 1Department of Medicine B, Sheba Medical Center, Tel Hashomer, Israel; 2The Dr. Pinchas Bornstein Talpiot Medical Leadership Program, Sheba Medical Center, Tel Hashomer, Israel; 3The Israel Defense Forces Medical Corps, Israel; 4The Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel; 5Division of Endocrinology and Metabolism, and Population Health Research Institute, McMaster University and Hamilton Health Sciences, Hamilton, Ontario, Canada; 6Department of Obstetrics and Gynecology, Sheba Medical Center, Tel Hashomer, Israel; 7 Israel Ministry of Health, Jerusalem, Israel; 8Department of Endocrinology, Sheba Medical Center, Tel Hashomer, Israel; 9Department of Psychiatry B, Ness Ziona Mental Health Center, Ness Ziona, Israel; and 10 Division of Endocrinology, Diabetes and Hypertension, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts Address correspondence to: Gilad Twig, MD, PhD, Department of Medicine B, Chaim Sheba Medical Center, Tel Hashomer, Ramat Gan 52621, Israel. E-mail: [email protected]. 0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2015.12.006
& 2015 American Journal of Preventive Medicine
T
he incidence of Type 2 diabetes is increasing worldwide, involving an increasing number of young adults. The observation that established risk factors such as obesity cannot solely account for this increase1,2 highlights the need to identify novel risk factors. Psychosocial factors, including emotional distress, have been suggested as potential risk factors for future diabetes.3–5 However, the data are conflicting, with some prospective studies reporting a twofold or higher risk of diabetes in people reporting excessive distress,6–8 whereas others report equivocal risk9,10 or even a risk reduction.11 Each of these studies defined emotional distress differently while utilizing various approaches to adjust for confounders, with limited consideration of distress level changes over time. Moreover, these studies
Published by Elsevier Inc.
Am J Prev Med 2015;](]):]]]–]]] 1
2
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
usually relied on a single baseline assessment in middleaged individuals in their sixth decade of life or older. The Metabolic, Lifestyle, and Nutrition Assessment in Young Adults (MELANY) cohort is an ongoing, large, prospective study in Israel assessing the effect of clinical and biochemical parameters on cardiometabolic morbidity among young adults.12–15 Data from more than 32,000 young healthy men were analyzed to explore the relationship between emotional distress and incident diabetes after accounting for a comprehensive set of confounders and after considering changes in distress level over time.
Methods The Metabolic, Lifestyle, and Nutritional Assessment in Young Adults Cohort The MELANY cohort is being conducted at the Israel Defense Forces screening center (Zrifin, Israel), to which all career service personnel aged 425 years are referred every 3–5 years for routine health examinations and screening tests.12,16 At each visit to the screening center, participants complete a detailed questionnaire in Hebrew assessing nutritional, medical, and lifestyle factors, including a brief question regarding self-perceived emotional distress. Blood samples are drawn and analyzed immediately following a 14-hour fast. Complete physical examination is performed by a physician at the center and includes measurements of blood pressure, weight, and height. Primary care for all Israel Defense Forces personnel between scheduled visits to the center is obtained at designated military clinics, and all medical information is recorded in a central computerized database.
Study Population Figure 1 depicts the study design. Between 1995 and 2011, all firsttime attendees examined at the screening center were asked the
following question: Are you preoccupied by worries or concerns that affect your overall wellbeing? This analysis included all men who answered either yes, no, or I am not sure; had a fasting plasma glucose (FPG) level o126 mg/dL; and denied a history of depression or diabetes. The IRB of the Israeli Defense Medical Corps approved this study with the assurance of strict participant anonymity during data analyses.
Measures Participants were followed prospectively from enrollment (first visit to the screening center) and the primary outcome was diabetes onset. Participants underwent the same biochemical and clinical evaluation (including screening for emotional distress and FPG-based screening for diabetes) at each consecutive visit. All subjects were censored at the time of death, retirement from military service, March 8, 2011, or diabetes diagnosis, whichever came first. Incident cases of diabetes were diagnosed by a physician according to the American Diabetes Association diagnostic criteria, by documenting either two FPG levels Z126 mg/dL or a glucose level Z200 mg/dL 2 hours after ingestion of 75 grams of glucose. No antibody data were available; as such, the type of diabetes (e.g., Type 1) could not be ascertained. However, it was recently reported for this cohort that 498% of diabetes diagnosed cases were not prescribed insulin during the first year, thereby supporting Type 2 predominance. All laboratory studies were performed on fresh samples in an ISO-9002 quality-assured core facility laboratory. Emotional distress was assessed by the question: Are you preoccupied by worries or concerns that affect your overall wellbeing? (yes, no, uncertain). The assessment of distress by similar single-item measures exhibited good test–retest reliability and good correlation with multi-item distress questionnaires,17 and was used to establish an association with different clinical outcomes,18–20 including diabetes.6 The Mini Sleep Questionnaire (MSQ) is a ten-question questionnaire used in clinical studies to assess sleep quality.21–23 Cognitive performance at pre-recruitment evaluation (approximately at age 17 years) was denoted by a
Figure 1. Diagram of study design and outcomes. www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]] 24
general intelligence score (GIS) as previously described. The GIS was shown to correlate well with the Wechsler Adult Intelligence Scale total IQ25 and was found to serve as an independent risk factor for diabetes in this cohort in earlier analyses.24,26 Both MSQ and GIS were modeled as continuous variables. SES data were coded by the area of residence and was stratified according to a 1– 10 scale for all municipalities in Israel as devised by the Israeli Central Bureau of Statistics.27,28 SES was categorized into the groups: low (1–4); medium (5–7); and high (8–10); as reported previously.27,28 Marital status (single, living with a spouse, previously married and currently living alone); consumption of sleeping or relaxation pills (no, yes); type of army service (field versus home front predominant); and education level (r8, r12, Z13 years of education) were treated as categorical variables. Country of origin (classified by the father’s or grandfather’s country of birth) was categorized into five geographic areas:13,24 former USSR countries; Asia (non-USSR); Africa (excluding South Africa); Western (comprised of non-USSR Europe, North and South America, South Africa, Australia, and New Zealand); and Israel. Birth country (indicating nativity rather than ancestry) was classified in a similar manner. Statistical analysis was conducted between 2014 and 2015. Age; triglyceride level; FPG; high-density lipoprotein cholesterol (HDL-c); and white blood cell count were treated as continuous variables. Physical activity (not active, o150 minutes/week, Z150 minutes/week); smoking status (never, ex-smoker, current); family history of diabetes (yes, no); regular consumption of alcoholic beverages (yes, no); and regular breakfast consumption (yes, no) were treated as categorical variables. BMI was treated as a continuous variable in multivariate analysis, and as a categorical variable (normal, BMI o25; overweight, 25r BMI o30; obese, BMI Z30) assessing an interaction between BMI and distress level with future diabetes risk.
Statistical Analysis Continuous variables were summarized using means and SDs or medians with intraquartile ranges. Counts with percentages were used for binary variables. Comparison of baseline characteristics was conducted using unpaired t-tests and chi-square tests for continuous and categorical variables, respectively. Cox regression models were used to estimate the hazard ratios (HRs) and 95% CIs for diabetes after adjustment for age alone (Model 1) and after adjustment for additional clusters of risk factors (Models 1A–1D), including genetic (age, country of origin, family history of diabetes); lifestyle-related (age, physical activity, smoking status, alcohol consumption, breakfast consumption); metabolic (age, BMI, FPG, HDL-c, WBC count, triglyceride level); and psychosocial (age, type of army service, SES, GIS, education, and MSQ score). Variables that were significant at po0.05 in the ageadjusted analyses were included in the final multivariable analysis. A forward likelihood ratio stepwise model was applied for the covariates listed in the final multivariate model to rank the contribution of emotional distress with respect to other modifiable risk factors. In this analysis, non-modifiable diabetes risk factors were first introduced (age, country of origin, family history of diabetes, cognitive performance) and then the contribution of the modifiable risk was determined (education, BMI, FPG, physical activity, triglyceride level, HDL-c level, MSQ score). Log minus log plots for each categorical variable were inspected to verify the ] 2015
3
assumption of proportionality of the hazards. The final multivariate model was also performed with terms of interaction between the continuous variables and log(time) (SAS, version 13.1, PROC PHREG). Distress status, BMI, smoking status, and physical activity were also treated as time-dependent variables as appropriate. Differences in follow-up according to degree of emotional distress were compared with ANOVA and Dunnet T3 post hoc multiple comparison tests. Two supplemental analyses were done. The first was limited to participants who attended the screening center at least twice, and defined diabetes based on an FPG value Z126 mg/ dL at scheduled visits only. This was done to account for the possibility of excess consumption of healthcare services by participants reporting distress. The second was limited to participants with a minimal follow-up duration of 6 years, to account for differential follow-up of participants with and without emotional distress. Participants with missing covariate data (final model, 13.7%) were excluded from multivariable analysis. A separate analysis applied imputation for missing values using a decision tree algorithm for categorical variables and distribution methods for continuous variables (SAS Miner, version 13.1). Values represent the mean (SD), unless mentioned otherwise. Analyses were performed with SPSS, version 21.0, unless stated otherwise.
Results Table 1 presents the baseline characteristics of the cohort. Emotional distress was reported by 1,341 participants (4.1%). This subpopulation was characterized by an increase in BMI (25.8 vs 25.4, p¼0.002) and higher obesity rates (15.7% vs 12.0%, po0.001); lower degree of physical activity; and higher degree of current (or past smoking) as compared with those denying distress. FPG, prevalence of impaired fasting glucose, HDL-c, and blood pressure were comparable at baseline between groups. Participants who reported emotional distress had a threefold increase in the prevalence of abnormal sleep score (MSQ score 424, 67% vs 22%). Accordingly, consumption of sleeping and relaxation pills was more than fivefold higher among those reporting emotional distress compared with those denying it (7.9% vs 1.4%, respectively). There were 723 new cases of diabetes diagnosed during 206,382 person-years of follow-up (mean total follow-up of 6.3 [4.3] years). The incident rate of diabetes among those denying any emotional distress was 3.32 cases/ 1,000 person-years, and was nearly doubled when emotional distress was reported (6.35 cases/1,000 personyears, Table 2). This difference persisted in various models adjusting for various genetic, lifestyle, and metabolic risk factors for diabetes (Appendix Figure 1, available online). There was an HR of 1.53 attributed to emotional distress (95% CI¼1.08, 2.18, p¼0.017, Table 2) in a multivariate model adjusted for age, BMI, FPG, family history of diabetes, triglyceride level, HDL-c,
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
4
Table 1. Baseline Characteristics of the Cohort at Enrollment (First Visit to the Screening Center) Are you preoccupied by worries or concerns that affect your overall wellbeing? Characteristic n Age (years)
No
Uncertain
Yes
Total
29,797
1,447
1,341
32,586
30.9⫾5.6
31.4⫾6.0
31.6⫾6.0
31.0⫾5.6
Country of origin
p-value
o0.001 0.112
Israel
8
9
8
8
Former USSR countries
12
14
17
12
Asia
24
24
23
24
Africa
31
32
31
31
West
25
21
21
25
Marital status
0.042
Single
21.2
24.8
22.5
21.4
Living with a spouse
76.7
72.9
71.7
76.3
Divorced, living alone
2.2
2.3
5.7
2.3
SES
0.091
Low
32
32
33
Medium
53
55
54
High
15
13
13
81.4
75.1
77.6
Z13 years of education
80.9
Type of service
0.006 0.070
Field-dominant
38.4
36.1
34.5
38.1
Desk-dominant
61.6
63.9
65.5
61.9
25.43⫾4.0
25.50⫾4.1
25.82⫾4.2
25.45⫾4.0
Normal weight
50
50
46.5
50
Overweight
38
36
37.8
38
Obese
12
14
15.7
12
89.5⫾9.0
89.6⫾9.1
89.4⫾9.3
89.5⫾9.0
0.742
11.9
13.1
11.5
11.9
0.345
117.9⫾12.5
117.6⫾12.4
117.8⫾12.2
117.5⫾12.5
0.171
BPDiastolic
73.6 ⫾9.6
74.6⫾ 10.2
73.9 ⫾10.0
74.1⫾9.5
0.569
HDL-c (mg/dL)
46.3⫾10.7
45.9⫾10.5
46.0⫾11.1
46.2⫾10.7
0.237
101 (71, 147)
99 (72, 152)
105 (74, 154)
101 (71, 148)
0.072
117.2⫾33.1
118.1⫾33.0
116.6⫾34.6
117.2⫾33.2
0.574
6.65⫾1.49
6.72⫾1.49
6.84⫾1.56
6.66⫾1.49
o0.001
BMI
Fasting glucose level (mg/dL) Impaired fasting glucose BPSystolic (mean ⫾ SD, mmHg)
Triglycerides (mg/dL) (25th, 75th) LDL (mg/dL) 3
3
WBC count (10 cells/mm )
0.002
o0.001
Physical activity None
64
74
78
65
o150 minutes/week
28
21
17
27
Z150 minutes/week
8
5
5
8 (continued on next page)
www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
5
Table 1. Baseline Characteristics of the Cohort at Enrollment (First Visit to the Screening Center) (continued) Are you preoccupied by worries or concerns that affect your overall wellbeing? Characteristic Family history of diabetes
No
Uncertain
Yes
Total
p-value
13.8
19.1
18.7
14.2
o0.001 o0.001
Smoking status Never
60
52
50
59
Ex-smoker
13
15
13
13
Current
27
33
37
28
Abnormal MSQ score
23
59
68
26
o0.001
Use of sleeping pills
1.4
7.9
5.9
1.9
o0.001
Breakfast consumption
80
85
83
81
0.030
Alcohol consumption
8.0
11.5
11.5
8.2
o0.001
Note: Boldface indicates statistical significance (po0.05). BP, blood pressure; HDL-c, high-density lipoprotein cholesterol; LDL, low-density lipoprotein cholesterol; MSQ, Mini Sleep Questionnaire; USSR, Union of Soviet Socialist Republic; WBC, white blood cell.
education, GIS, white blood cell count, physical activity, and MSQ score. In a forward stepwise model, emotional distress was introduced to the model in the sixth step among modifiable risk factors, preceded by FPG, BMI, MSQ score, education, and triglyceride level (change in
chi square, 7.489; p¼0.024). Figure 2A depicts the cumulative diabetes incidence in the group with reported emotional distress at baseline versus those without. Similar results were obtained when the definition of diabetes was limited to those with an FPG Z126 mg/dL
Table 2. Hazard Ratio (HR) for Developing Diabetes Depending on Baseline Distress Level Are you preoccupied by worries or concerns that affect your overall wellbeing? Variable
No
Uncertain
Yes
Total
29,797
1,447
1,341
32,586
New cases of diabetes
635
42
46
723
Mean follow-up (years)
6.41⫾4.29
5.66⫾4.18
5.40⫾4.030
6.33⫾4.28
190,943
8,190
7,239
206,382
3.32
5.12
6.35
3.50
1 (ref)
1.624
1.899
95% CI
1.188, 2.221
1.407, 2.562
p-value
0.002
o0.001
n
Cumulative follow-up (person-years) Rate (1/1,000 person-years) Model 1: Age HR
Model 2: Age, BMI, FPG, family history of diabetes, triglyceride level, HDL-c, education, GIS, WBC, physical activity, MSQ sleep quality score HR
1 (ref)
1.361
1.530
95% CI
0.789, 1.676
1.081, 2.183
p-value
0.467
0.017
FPG, fasting plasma glucose; GIS, general intelligence score; HDL-c, high-density lipoprotein cholesterol; HR, hazard ratio; MSQ, Mini Sleep Questionnaire; WBC, white blood cells.
] 2015
6
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
Figure 2. Emotional distress level over time and diabetes incidence. Note: Incidence of cumulative diabetes over time for a single evaluation (at enrollment) and for two consequent evaluations of emotional distress level (a and b respectively). Diabetes risk was adjusted for the multivariable model indicated in Table 2 (model 2). Detailed hazard ratio and follow-up data for (a) are shown in Table 2, and for (b) in Appendix Table 2 (available online).
at scheduled follow-up visits (n¼242 cases of diabetes; occurrence of emotional distress, HR¼1.85, 95% CI¼1.19, 2.87, p¼0.006), or when only participants with a follow-up of 46 years were included (339 of 12,831 participants, HR¼1.92, 95% CI¼1.22, 3.01, p¼0.005). Of note, participants reporting emotional distress had a higher risk for diabetes than those denying it despite a shorter follow-up duration (po0.001, Table 2). Similar results were obtained for Models 1A–1D and Model 2 when imputation for missing variables was applied, or when interaction between continuous variables and log (time) was introduced into the model (data not shown). As noted in Appendix Table 1 (available online), the ageadjusted diabetes incidence of 2,914 men who did not provide an answer was similar to those who did reply. Increased body weight has been suggested as a potential link between psychosocial distress and increased diabetes risk.29 However, no attenuation in the distress–diabetes association was found when BMI was treated as a time-dependent variable (HR [Model 2]¼1.70, 95% CI¼1.21, 2.38, p¼0.002). There was no interaction between BMI and level of emotional distress with future diabetes incidence (p¼0.393). Data regarding emotional distress at two time points were available for 10,709 participants of the study population, with a minimal time interval of 2 years between scheduled visits and a subsequent follow-up of 9.50 (3.3) years (median, 9.99 years). After two consecutive visits, 8,935 participants (83.4%) consistently denied emotional distress (no–no), whereas 117 participants (1.1%) consistently confirmed it (yes–yes). Of the
442 newly diagnosed cases of diabetes, 3.76 cases/1,000 person-years occurred in the no–no group, whereas 7.69 cases/1,000 person-years occurred in the yes–yes group. This difference was translated to an adjusted HR of 2.14 (Model 2, 95% CI¼1.04, 4.47, p¼0.041) in those reporting sustained emotional distress compared with those consistently denying it (Figure 2B). Diabetes risk was not statistically different among individuals reporting decrease in emotional distress level (yes–no) and those that denied distress (Model 2, HR¼0.92, 95% CI¼0.43, 1.97, p¼0.836). When the level of emotional distress was treated as a time-dependent variable, an HR of 1.70 (95% CI¼1.25, 2.31, p¼0.001) was obtained for incident diabetes, compared with cases where it was absent. The distress–diabetes association remained significant when BMI, physical activity, and smoking status were treated as time-dependent variables (Model 2, HR¼1.66, 95% CI¼1.21, 2.17, p¼0.002). Of note, the age of diabetes onset among participants with two consequent evaluations compared with those with one evaluation was mildly delayed (41.1 [3.9] years vs 37.8 [5.0] years, po0.001).
Discussion This analysis of 32,586 young men followed for more than 200,000 person-years demonstrates an association between self-perceived emotional distress and incident diabetes. Emotional distress was found in 4.1% of the cohort and increased the risk for incident diabetes by 1.53-fold, independent of other known clinical and www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
biochemical diabetes risk factors. When emotional distress persisted in two consecutive visits, cumulative diabetes incidence exceeded 15%, whereas normalization of the distress level restored the diabetes incidence to levels similar to those without distress. Accordingly, emotional distress level was also found to be a significant risk factor in time-dependent analysis. A variety of mechanisms may mediate the distress– diabetes association. Psychological distress may dysregulate the nervous system and hypothalamic–pituitary axis, resulting in increased levels of cortisol, adrenaline, and neuropeptide Y.30,31 Abnormally elevated levels of these stress hormones trigger gluconeogenesis and lipolysis while suppressing insulin secretion and decreasing insulin sensitivity.32,33 Furthermore, distress is associated with low-grade chronic inflammatory response that can further aggravate insulin resistance.30 The observational nature of this study precludes characterization of these mechanisms. The present analyses have unique strengths. As opposed to previous studies,34–36 metabolism-related risk factors including time-dependent effect of obesity status and smoking were considered,37 consistent with a recent large study.5 Also, the authors assessed the continuous effect of emotional distress in a time-dependent manner rather than a single baseline assessment.5,29,36,37 In order to address the possibility that emotional distress may increase healthcare service use, thereby increasing the likelihood of diabetes diagnosis (sampling bias), analyses restricted to diabetes diagnosis defined as FPG Z126 mg/ dL measured at scheduled visits to the screening center were performed. Additionally, this study examined the possibility that low SES38 may confound the link between emotional distress and incident diabetes. However, the distress–diabetes relationship persisted after accounting for socioeconomic-related covariates (Table 2; Appendix Figure 1, available online). Finally, although the relationship between depressive symptoms and disturbed glucose levels may be bidirectional,35 it is unlikely that higher levels of emotional distress were due to undiagnosed diabetes at baseline. This is supported by the comparable glucose level at baseline in those with and without emotional distress and the persistence of the distress–diabetes association even after a 6-year “wash-out” period.
Limitations The limitations of this study include the use of selfreported, rather than objective, data to assess for emotional distress level. Also, evaluation of distress by an implicit question, as used in the current and previous reports,6,7 may suffer from suboptimal sensitivity. Nevertheless, single-item measures of emotional distress are ] 2015
7 17
straightforward with good test–retest reliability, easy to use in various clinical settings,39–41 and correlate well with multi-item questionnaires assessing distress.17 Additionally, the association between distress and a variety of other clinical metabolic outcomes, such as ischemic heart disease has also been established using single-item measures.18–20 Second, the MELANY cohort is restricted to a unique group of army recruits. Yet, the current findings were independent of the type of military service (Appendix Figure 1, available online), and the characteristics of the participants of the MELANY cohort were highly similar to other U.S.- and European-based cohorts of young men42–44 and are considered to adequately reflect the population of young adults in Israel.28 The characteristics and design of our cohort are also advantageous in minimizing the effect of socioeconomic confounders and by including repetitive measurements of clinical and metabolic variables with no loss to follow-up.24 Third, the analysis was conducted only in men, thereby limiting the generalization of the results to women. Finally, the distress–diabetes association is complex and may be accompanied by additional confounders that were not, or only partially, adjusted for. Nevertheless, the applied multivariate model (Table 2) represents a detailed clinical and biochemical assessment that is acceptable in clinical practice.
Conclusions This study demonstrates an association between selfperceived emotional distress over time and increased risk for diabetes among apparently young healthy men. These results emphasize the need for awareness by physicians when stratifying the risk for diabetes. This study was supported by a research grant of the Israel Defense Forces Medical Corp and the Israeli ministry of defense. GT and AT were partially supported by a grant from the Pinchas Borenstein Talpiot Medical Leadership Program, Sheba Medical Center, Tel Hashomer, Israel. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions included the following: study concept and design (GT); data acquisition (DT); statistical analyses (GT, ED); interpretation of data (GT, HCG, AS, ED, HA, EF, TCY, AT); writing the first draft of the manuscript (GT); and critical revision of the manuscript (AA, AS, AT, ED, DA, TCY, HCG, EF). The content of this manuscript was presented in the 75th American Diabetes Association meeting (Boston, MA). GT had full access to all the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis. No financial disclosures were reported by the authors of this paper.
8
Twig et al / Am J Prev Med 2015;](]):]]]–]]]
References 1. Menke A, Rust KF, Fradkin J, Cheng YJ, Cowie CC. Associations between trends in race/ethnicity, aging, and body mass index with diabetes prevalence in the United States: a series of cross-sectional studies. Ann Intern Med. 2014;161(5):328–335. http://dx.doi.org/ 10.7326/M14-0286. 2. Lyssenko V, Laakso M. Genetic screening for the risk of type 2 diabetes: worthless or valuable? Diabetes Care. 2013;36(suppl 2):S120– S126. http://dx.doi.org/10.2337/dcS13-2009. 3. Chandola T, Brunner E, Marmot M. Chronic stress at work and the metabolic syndrome: prospective study. BMJ. 2006;332(7540):521–525. http://dx.doi.org/10.1136/bmj.38693.435301.80. 4. Bergmann N, Gyntelberg F, Faber J. The appraisal of chronic stress and the development of the metabolic syndrome: a systematic review of prospective cohort studies. Endocr Connect. 2014;3(2):R55–R80. http: //dx.doi.org/10.1530/EC-14-0031. 5. Virtanen M, Ferrie JE, Tabak AG, et al. Psychological distress and incidence of type 2 diabetes in high-risk and low-risk populations: the Whitehall II Cohort Study. Diabetes Care. 2014;37(8):2091–2097. http://dx.doi.org/10.2337/dc13-2725. 6. Rod NH, Gronbaek M, Schnohr P, Prescott E, Kristensen TS. Perceived stress as a risk factor for changes in health behaviour and cardiac risk profile: a longitudinal study. J Intern Med. 2009;266(5):467–475. http://dx.doi.org/10.1111/j.1365-2796.2009.02124.x. 7. Novak M, Bjorck L, Giang KW, Heden-Stahl C, Wilhelmsen L, Rosengren A. Perceived stress and incidence of Type 2 diabetes: a 35-year follow-up study of middle-aged Swedish men. Diabet Med. 2013;30(1):e8–e16. http://dx.doi.org/10.1111/dme.12037. 8. Eriksson AK, Ekbom A, Granath F, Hilding A, Efendic S, Ostenson CG. Psychological distress and risk of pre-diabetes and Type 2 diabetes in a prospective study of Swedish middle-aged men and women. Diabet Med. 2008;25(7):834–842. http://dx.doi.org/10.1111/j.1464-5491.2008.02463.x. 9. Williams ED, Magliano DJ, Tapp RJ, Oldenburg BF, Shaw JE. Psychosocial stress predicts abnormal glucose metabolism: the Australian Diabetes, Obesity and Lifestyle (AusDiab) study. Ann Behav Med. 2013;46(1):62–72. http://dx.doi.org/10.1007/s12160-013-9473-y. 10. Mommersteeg PM, Herr R, Zijlstra WP, Schneider S, Pouwer F. Higher levels of psychological distress are associated with a higher risk of incident diabetes during 18 year follow-up: results from the British Household Panel Survey. BMC Public Health. 2012;12:1109. http://dx. doi.org/10.1186/1471-2458-12-1109. 11. Eriksson AK, van den Donk M, Hilding A, Ostenson CG. Work stress, sense of coherence, and risk of type 2 diabetes in a prospective study of middle-aged Swedish men and women. Diabetes Care. 2013;36(9): 2683–2689. http://dx.doi.org/10.2337/dc12-1738. 12. Tirosh A, Shai I, Tekes-Manova D, et al. Normal fasting plasma glucose levels and type 2 diabetes in young men. N Engl J Med. 2005;353(14): 1454–1462. http://dx.doi.org/10.1056/NEJMoa050080. 13. Twig G, Afek A, Derazne E, et al. Diabetes risk among overweight and obese metabolically healthy young adults. Diabetes Care. 2014;37(11): 2989–2995. http://dx.doi.org/10.2337/dc14-0869. 14. Twig G, Gerstein HC, Ben-Ami Shor D, et al. Coronary artery disease risk among obese metabolically healthy young men. Eur J Endocrinol. 2015;173(3):305–312. http://dx.doi.org/10.1530/EJE-15-0284. 15. Furer A, Afek A, Beer Z, et al. Height at late adolescence and incident diabetes among young men. PLoS One. 2015;10(8):e0136464. http://dx. doi.org/10.1371/journal.pone.0136464. 16. Tirosh A, Shai I, Afek A, et al. Adolescent BMI trajectory and risk of diabetes versus coronary disease. N Engl J Med. 2011;364(14):1315– 1325. http://dx.doi.org/10.1056/NEJMoa1006992. 17. Littman AJ, White E, Satia JA, Bowen DJ, Kristal AR. Reliability and validity of 2 single-item measures of psychosocial stress. Epidemiology. 2006;17(4):398–403. http://dx.doi.org/10.1097/01.ede.0000219721. 89552.51.
18. Nielsen NR, Kristensen TS, Prescott E, Larsen KS, Schnohr P, Gronbaek M. Perceived stress and risk of ischemic heart disease: causation or bias? Epidemiology. 2006;17(4):391–397. http://dx.doi. org/10.1097/01.ede.0000220556.86419.76. 19. Nielsen NR, Kristensen TS, Schnohr P, Gronbaek M. Perceived stress and cause-specific mortality among men and women: results from a prospective cohort study. Am J Epidemiol. 2008;168(5):481–491; discussion 492–486. 10.1093/aje/kwn157. 20. Non AL, Rimm EB, Kawachi I, Rewak MA, Kubzansky LD. The effects of stress at work and at home on inflammation and endothelial dysfunction. PLoS One. 2014;9(4):e94474. http://dx.doi.org/10.1371/ journal.pone.0094474. 21. Falavigna A, de Souza Bezerra ML, Teles AR, et al. Sleep disorders among undergraduate students in Southern Brazil. Sleep Breath. 2011;15(3):519–524. http://dx.doi.org/10.1007/s11325-010-0396-6. 22. Melamed S, Oksenberg A. Excessive daytime sleepiness and risk of occupational injuries in non-shift daytime workers. Sleep. 2002;25(3): 315–322. 23. Twig G, Shina A, Afek A, et al. Sleep quality and risk of diabetes and coronary artery disease among young men. Acta Diabetol. 2015. http: //dx.doi.org/10.1007/s00592-015-0779-z. 24. Twig G, Gluzman I, Tirosh A, et al. Cognitive function and the risk for diabetes among young men. Diabetes Care. 2014;37(11):2982–2988. http://dx.doi.org/10.2337/dc14-0715. 25. Reichenberg A, Weiser M, Rabinowitz J, et al. A population-based cohort study of premorbid intellectual, language, and behavioral functioning in patients with schizophrenia, schizoaffective disorder, and nonpsychotic bipolar disorder. Am J Psychiatry. 2002;159(12): 2027–2035. http://dx.doi.org/10.1176/appi.ajp.159.12.2027. 26. Cukierman-Yaffe T, Kasher-Meron M, Fruchter E, et al. Cognitive performance at late adolescence and the risk for impaired fasting glucose among young adults. J Clin Endocrinol Metab. 2015;100 (12):4409–4416. http://dx.doi.org/10.1210/jc.2015-2012. 27. Twig G, Afek A, Shamiss A, et al. Adolescence BMI and trends in adulthood mortality: a study of 2.16 million adolescents. J Clin Endocrinol Metab. 2014;99(6):2095–2103. http://dx.doi.org/10.1210/jc.2014-1213. 28. Twig G, Livneh A, Vivante A, et al. Cardiovascular and metabolic risk factors in inherited autoinflammation. J Clin Endocrinol Metab. 2014;99(10):E2123–E2128. http://dx.doi.org/10.1210/jc.2014-2096. 29. Heraclides AM, Chandola T, Witte DR, Brunner EJ. Work stress, obesity and the risk of type 2 diabetes: gender-specific bidirectional effect in the Whitehall II study. Obesity (Silver Spring). 2012;20(2):428– 433. http://dx.doi.org/10.1038/oby.2011.95. 30. McEwen BS. Protective and damaging effects of stress mediators. N Engl J Med. 1998;338(3):171–179. http://dx.doi.org/10.1056/NEJM 199801153380307. 31. Bjorntorp P. Body fat distribution, insulin resistance, and metabolic diseases. Nutrition. 1997;13(9):795–803. http://dx.doi.org/10.1016/S08999007(97)00191-3. 32. Lundberg U. Stress hormones in health and illness: the roles of work and gender. Psychoneuroendocrinology. 2005;30(10):1017–1021. http: //dx.doi.org/10.1016/j.psyneuen.2005.03.014. 33. Bhathena SJ. Relationship between fatty acids and the endocrine and neuroendocrine system. Nutr Neurosci. 2006;9(1-2):1–10. http://dx. doi.org/10.1080/10284150600627128. 34. Knol MJ, Twisk JW, Beekman AT, Heine RJ, Snoek FJ, Pouwer F. Depression as a risk factor for the onset of type 2 diabetes mellitus. A meta-analysis. Diabetologia. 2006;49(5):837–845. http://dx.doi.org/ 10.1007/s00125-006-0159-x. 35. Mezuk B, Eaton WW, Albrecht S, Golden SH. Depression and type 2 diabetes over the lifespan: a meta-analysis. Diabetes Care. 2008; 31(12):2383–2390. http://dx.doi.org/10.2337/dc08-0985. 36. Nyberg ST, Fransson EI, Heikkila K, et al. Job strain as a risk factor for type 2 diabetes: a pooled analysis of 124,808 men and women. Diabetes Care. 2014;37(8):2268–2275. http://dx.doi.org/10.2337/dc13-2936.
www.ajpmonline.org
Twig et al / Am J Prev Med 2015;](]):]]]–]]] 37. Mooy JM, de Vries H, Grootenhuis PA, Bouter LM, Heine RJ. Major stressful life events in relation to prevalence of undetected type 2 diabetes: the Hoorn Study. Diabetes Care. 2000;23(2):197–201. http: //dx.doi.org/10.2337/diacare.23.2.197. 38. Kumari M, Head J, Marmot M. Prospective study of social and other risk factors for incidence of type 2 diabetes in the Whitehall II study. Arch Intern Med. 2004;164(17):1873–1880. http://dx.doi.org/10.1001/ archinte.164.17.1873. 39. Hermelink K, Hohn H, Hasmuller S, et al. Brief distress screening in clinical practice: does it help to effectively allocate psycho-oncological support to female cancer inpatients? Breast Care (Basel). 2014;9(2): 129–133. http://dx.doi.org/10.1159/000360788. 40. Schubart JR, Mitchell AJ, Dietrich L, Gusani NJ. Accuracy of the Emotion Thermometers (ET) screening tool in patients undergoing surgery for upper gastrointestinal malignancies. J Psychosoc Oncol. 2015;33(1):1–14. http://dx.doi.org/10.1080/07347332.2014. 977415. 41. Keir ST, Calhoun-Eagan RD, Swartz JJ, Saleh OA, Friedman HS. Screening for distress in patients with brain cancer using the NCCNʼs
] 2015
9
rapid screening measure. Psychooncology. 2008;17(6):621–625. http: //dx.doi.org/10.1002/pon.1271. 42. Vassy JL, Hivert MF, Porneala B, et al. Polygenic type 2 diabetes prediction at the limit of common variant detection. Diabetes. 2014;63 (6):2172–2182. http://dx.doi.org/10.2337/db13-1663. 43. Wurtz P, Soininen P, Kangas AJ, et al. Branched-chain and aromatic amino acids are predictors of insulin resistance in young adults. Diabetes Care. 2013;36(3):648–655. http://dx.doi.org/10.2337/dc12-0895. 44. Nguyen QM, Srinivasan SR, Xu JH, et al. Elevated liver function enzymes are related to the development of prediabetes and type 2 diabetes in younger adults: the Bogalusa Heart Study. Diabetes Care. 2011;34(12):2603–2607. http://dx.doi.org/10.2337/dc11-0919.
Appendix Supporting data Supplementary data associated with this article can be found at http://dx.doi.org/10.1016/j.amepre.2015.12.006.