Stress research over a ten year period what have we ... · Stress research over a ten year period – what have we learned, trends and ... depressive episode Psychosocial factors

Stress research over a ten year period –what have we learned, trends and

future?

Professor Mika KivimakiDepartment of Epidemiology & Public Health University

College London, UK

Session Outline

Stress, stressors and the disease process

As a trigger of cardiac event among vulnerable individuals

Chronic stress and development of diabetes evidence?

Next steps

What are the sources of psychological stress?

A. Negative thought patterns and emotions1. Depressive symptoms and syndromes2. Anxiety symptoms and syndromes3. Hostility and anger4. Rumination5. Resentment6. Pessimism

C. Unsatisfied basic psychological needs1. Lack of sense of purpose2. Lack of social connectness3. Lack of sense of security4. Lack of autonomy

B. Chronic stress1. Work stress2. Marital stress and dissatisfaction3. Social isolation, lack of social support4. Caregiver strain5. Trauma or abuse in childhood6. Perceived injustice

D. Lack of rest and relaxation1. Sleep loss2. Difficulty in unwinding3. Time pressure

Rozanski . In Hjelmdahl et al. 2012

PART 1:CAUSAL MODELS

Risk factors x

Preclinical disease processes

Cardiovascular death

Manifest disease x

e.g., obesity, smoking, physical inactivity

e.g., atherosclerosis, hyperglycaemia

e.g., cardiac event, depressive episode

Psychosocial factors

indirect effect prognostic factordirect effect trigger

PART 2:ALTERNATIVEEXPLANATIONS Psychosocial

factors stress

confounding, bias, reversed causality

Risk factors x


Manifest disease x

Complications, death

Psychologicalstress

Psychological stress

Stress and disease

Kivimäki et al. Scand J Work Environ Health 2006

Stress as a trigger

• Major earthquake in Athens in 1981 ‐> an excess of cardiac deaths over the following 3 days

• Major industrial accidents, wars, terrorist attacks‐> increase in rates of acute MI/sudden cardiac death

• Major sport events

Risk ratio of death from MI or stroke = 1.4 (95%CI 1.1‐2.1) among Dutchmen aged >45 years on the day on which the Dutch team lost the Frenchunder dramatic corcumstances in the 1996 European Cup.

2.7‐fold (95%CI 2.3‐3.4) higher use of emergencies in the Munich region of Germany during the 2006 soccer World Cup on days on which Germanteam played, particularly in the 2h after the start of matches.

Steptoe & Kivimaki. Nature Reviews 2012

Unemployment rate (%) in Finland

Statistics Finland, Eurostat

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

18.0

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

Une

mpl

oym

ent (

%)

Finland EU-15

Short- and long-term associations of downsizing on cardiovascular mortality

*Adjusted for age, sex, socioeconomic status and type of employment.

First 4 years of the follow-up

Next four years of the follow-up

Extent of downsizingNo 1.0 1.0Intermediate 4.3 (1.3-14.7) 1.1 (0.5-2.1)High 5.1 (1.4-19.3) 1.4 (0.6-3.1)

Vahtera et al. Br Med J 2004

Theoretical models on chronic work stress

Karasek 1979Karasek & Theorell 1990

Job Strain Model

Job motivation and job stressmodels

J. S. Adams: Equity Theory on job motivation 1963

effortreward

J. Siegrist: Effort‐Reward Imbalance model 1996

What I put in my work What I get from my work

Organizational justice theory― 3 forms of justice perceptions

Distributive justice: fairness of outcomes (equity, equality, and needs)

Procedural justice: fairness of the methods or procedures used (decision criteria, voice, control of the process)

Relational justice: fairness of the interpersonal treatment received (dignity and respect)

Moorman 2001, Greenberg & Cropanzano 2001, Kivimaki et al Arch Intern Med 2005

Organisational justice questionnaire items

Decisions…• are well‐informed,• are consistently applied (the rules are applied equally for

everyone).

Management… • listens to the concerns of all those affected by the decision, • provides opportunities to appeal against or challenge the

decision,• tries to deal with us in a truthful manner.

Elovainio et al Am J Public Health 2002; Kivimäki et al Psychol Med 2003

Fairness/justice principle

Fundamental human behavior?

the brown capuchin monkey(Cebus apella)

Brosnan & de Waal. Nature 2003

Experiment:

Monkeys were tought in pairs to give a token (a white PVC pipe) to the experimenter to receive a food reward (high valued and low valued).

The equity test (ET): both partner exchanged token for a low‐value reward

The inequity test (IT): the partner got a high‐value and the subject a low‐value reward

The effort control test (EC): the partner got a high‐value reward for free and subject a low‐value reward

The food control test (FC): high‐value reward shown and low‐value reward given to the partner

Brosnan & de Waal. Nature 2003

Work stress and CHD: Meta‐analysis of prospective cohort studies published until 2006

Stress model Age‐ and sex‐adjusted

Multiple adjusted**

Job strain (10 studies) 1.45 (1.15 to 1.84) 1.11 (0.91 to 1.35)

Effort‐reward imbalance (4) 2.52 (1.63 to 3.90) 2.51 (1.58 to 3.98)

Organizational injustice (2) 1.62 (1.24 to 2.13) 1.47 (1.12 to 1.95)

Kivimäki et al. Scand J Work Environ Health 2006

*summary estimates**risk factors and potential mediatorsUpdated meta‐analysis including 15 cohort studies published until 2011

replicated the effect for job strain: Pooled hazard ratio = 1.39!

Steptoe & Kivimäki. Nature Reviews 2012

• People who worked 11 hours or more a day were 67% more likely to have a heart attack than those who worked shorter hours.

• When the normal measures that doctors use to assess someone's risk of heart disease were adjusted to take account of this finding (blood pressure, lipids, smoking, diabetes, age & sex), the resulting predictions were far more accurate ‐ an improvement of around 5 per cent

• This is equivalent to around 6,000 of the 125,000 people who suffer heart attacks in the UK each year.

Kivimaki et al. Ann Intern Med 2011

Hard work won't kill you? Well it might actuallyItisoftensaidthat"hardworkwon'tkillyou".

Monday April 4, 2011

Long hours at work may boost heart-attack riskBy Amanda Gardner, Health.com April 4, 2011 -- Updated 2143 GMT (0543 HKT)

UKNEWSHeart risk of long hours

Long hours at work increase heart risk Tue Apr 5, 2011

April 5, 2011

Kivimäki et al. Ann Intern Med 2011

A look at evidence

Stress diabetes

• High‐income countries– Ischaemic heart disease (15.8%)– Cerebrovascular disease (9.0%)– Lung cancer (5.1%)– Diabetes mellitus (4.8%)– COPD (4.1%)

• Low‐income countries– Ischaemic heart disease (13.4%)– HIV/AIDS (13.2%)– Cerebrovascular disease (8.2%)– COPD (5.5%)– Lower respiratory infections (5.1%)

Leading Causes of Death: WHO 2030 scenario

Mathers et al. PLoS Med 2006

According to an American Diabetes Association expert panel, up to 70% of individuals withprediabetes will eventually develop diabetes.

In a Chinese diabetes prevention trial, the 20‐year cumulative incidence of diabetes in controls withimpaired glucose tolerance defined with repeatedOGTTs was even higher (>90%)

Prediabetes ‐‐ a high‐risk status for diabetes

Tabak,…Kivimaki. Lancet 2012 (in press)

Diagnostic criteria for prediabetes

Tabak et al. Lancet 2009

Tabak,…Kivimaki. Lancet 2012 (in press)


Risk factors x



Manifest disease x

e.g., obesity, physical inactivity

e.g., intermediate glycemia/ prediabetes (IFG or IGT)


indirect effect prognostic factoretiological factor trigger


factors stress


Risk factors x


Manifest type 2 diabetes x




Psychosocial factors and type 2 diabetes

Stress diabetes‐related riskfactors

Phases of the Whitehall II study Phase Dates Type Participants

1 1985-1988 Screening / questionnaire 10,308

2 1989-1990 Questionnaire 8,132


4 1995-1996 Questionnaire 8,628


6 2001 Questionnaire 7,355


8 2006 Questionnaire 7,180


Questionnaire Clinical examinationDemographic datasocio-economic data (income + work change) retirementwork psychosocial factorsnon-work psychosocial factors (financial insecurity, control at home, family relationships)social engagement

Health behavioursSmoking, alcoholDiet - food frequency, physical activity

CVD & DiabetesDetails of CVD symptoms, investigations & treatment2-h Oral Glucose Tolerance Test (OGTT)

General HealthLongstanding illnessHospital admissionsMedicationsSF-36

Mental healthGHQ (anxiety, depression)CESD depression scaleSF-36, Activities of daily living (ADL), Instrumental ADL

Functioningwalking speed, chair stands, hand grip strengthbalance test, spirometry (peak expiratory vol)weight, height, waist hip ratio, BP

Neuroendocrineheart rate variability hypothalamic-pituitary-adrenal axis measurements (salivary cortisols)

Subclinical CVDECG: Minnesota codes, left ventricular massUltrasound carotid IMT (artery wall thickness) Endothelial function & flow mediated dilation (subset)

Lipidstotal + HDL cholesteroltriglycerides

Carbohydrate metabolismHbA1c, fasting and post load glucose and insulin

Genes50k CHIP, Metabochipserology, CRP, IL-6

Cognitive functionAH4, Mill Hill, memory, verbal fluency, MMSE

Chandola et al. Eur Heart J 2008

Meta‐analysis of stress as a predictor of BMI

Wardle et al. Obesity 2010

Longitudinal association between stress and weight change by baseline BMIThe Whitehall II study

Kivimaki et al. Int J Obesity 2006

IPD‐Work analysis

Nyberg,… Kivimaki for IPD‐WorkJ Intern Med 2011 [epub ahead of print]

Nyberg,… Kivimaki for IPD‐WorkJ Intern Med 2011 [epub ahead of print]

Stress prediabetes

Chronic stress and the metabolic syndrome in Whitehall II

Chandola et al. BMJ 2006

Chronic stress and the metabolic syndrome in Whitehall II

*Metabolic syndrome at phase 5. Exposure to stress was assessed at phases 1, 2, 3, 5 (a total of 10 years).


Chronic stress and the metabolic syndrome in Whitehall II by sex

*Metabolic syndrome at phase 5. Exposure to stress was assessed at phases 1, 2, 3, 5.


Evidence of an association between• stress and diabetes risk factors (physical inactivity and obesity)

• stress and prediabetic conditions, such as the metabolic syndrome

The association between stress and diabetes should therefore be plausible,

but does the evidence to date confirm this?Stress diabetes

Kawakami et al. JECH 1999

Work stress and type 2 diabetes in 2600 Japanese men

Work stress and prediabetes in 1300 Japanese male officeworkers

Nakanishi et al. Occup Environ Med 2001

Work stress and type 2 diabetes in 1300 Japanese male officeworkers

Nakanishi et al. Occup Environ Med 2001

Kroenke et al. Am J Epidemiol 2007

Nurses Health Study, USA

Nurses Health Study, USA

Kroenke et al. Am J Epidemiol 2007

Heraclides et al. Diabetes Care 2009*BUT in men 1.07 (95% CI 0.71‐1.63) !

Despite evidence of the association between• stress and diabetes risk factors (physical inactivity and obesity)

• stress and prediabetic conditions, such as the metabolic syndrome

The evidence of an association between stressand diabetes is inconsistent


Risk factors x



Manifest disease x

e.g., obesity, physical inactiv ity





factors stress


Risk factors x




Perceivedstress

Perceivedstress


Stress

Elevated diabetes risk factors

Metabolic syndrome/ prediabetes

Reduced depression and stress

Effect of stress on diabetesmasked

Time

Masked effect?

Prediabetes reduces risk of depressive symptoms and thus perceived stress?

Golden et al. JAMA 20085200 US men and women aged 45 to 84

Kivimaki et al. Diabetes Care 2009

N = 4228 men and women aged 50‐74

Not strong support for antidepressive effects of prediabetes in Whitehall II

Gale et al. Biol Psychiatry 2010

A linear association in 4300 US Veterans at mean age 39

HbA1c and probability of depression: The English Longitudinal Study of Ageing (ELSA)

Hamer et al. Psychosom Med 2010N = 4338, general population, mean age 63 years.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

3 4 5 6 7 8

HbA1c

Pro

babi

lity

of C

ES

D d

epre

ssio

n

A linear association in >1 Million Korean adults

Batty et al. JECH 2012 epub

Stress

Use of antidepressants

Diabetogenic effects of antidepressants

Effect of stress on diabetesinflated

Time

Inflated effect?

Does antidepressant medication increase the risk of type 2 diabetes?

• The Pharmo database, N = 204,034 (Knol et al. Diabetologia2008)

• The UK General Practice Research Database Study, N = 11,206 (Andersohn et al. Am J Psychiatry 2009)

• The Diabetes Prevention Program Trial, N = 3187 (Rubin et al. Diabetes Care 2008 and 2010)

• The Finnish Public Sector study, N = 5085 (Kivimaki et al. Diabetes Care 2010)

Detection bias

• The notion that people under the care of a physician for depression are more likely to have other hidden health problems (such as diabetes) diagnosed, compared with their untreated counterparts who have less regular contact with medical services.

Detection bias?

Kivimaki et al. Biol Psychiatry 2011

Kivimaki et al. Biol Psychiatry 2011

Antidepressant use not related to changes in glycaemia


Risk factors x



Manifest disease x

e.g., obesity, physical inactiv ity





factors stress


Risk factors x







Psychological distress as a predictor of mortality in 11,500 non‐diabetic and diabetic adults

Hamer et al. PsychosomMed 2010

Deaths/N Hazard ratio (95% CI)

Health Surveys for England and Scotland


Risk factors x


Cardiov ascular death

Manifest disease x

Physical inactivity(obesity)

Metablolic syndrome




factors stress


Risk factors x






Conclusions based on updated evidence:

Conclusions

‐ Association with depression repeatedly shown and likely to be true

‐ with CHD established, but might be imprecise due to publication bias and reverse causation bias

‐ with diabetes inconsistent, but few well‐powered studies with OGTT‐basedascertainment of diabetes available

‐ with stroke uncertain, but studies on triggering rare

‐ with cancer unclear

‐ Next steps to strengthen evidence: ‐Individual‐level meta‐analysis (‐>reverse causation, publication bias, riskgroups)‐Randomised controlled trial (‐>reversibility, confounding, bias)‐Instrumental‐regression analyses (‐>subjectivity bias)‐Mechanistic studies (‐>population‐level, chronic stress)

Stress research over a ten year period what have we ... · Stress research over a ten year period – what have we learned, trends and ... depressive episode Psychosocial factors

Documents