Background
There is a lack of attention paid to the potential role of psychosocial risk factors, including perceived psychosocial stress, in the development of stroke [
1,
2]. The association between psychosocial stress and the development of coronary heart disease is strong [
3‐
5]. A recently published overview of systematic reviews confirms modest to moderate evidence of the association between psychosocial stress at work and cardiovascular outcomes [
6]. Research is less conclusive in the area of stroke, yet public perception highlights psychosocial stress as a key risk factor for stroke [
7‐
9]. Several observational studies have identified an association between perceived psychosocial stress and stroke, [
10‐
13] however conflicting findings have been reported and inconsistencies are apparent with respect to definition and measurement of perceived psychosocial stress, study design and quality, duration of follow-up and number of covariates adjusted for. This variability has resulted in different, and sometimes contradictory, conclusions being drawn about the relationship between perceived psychosocial stress and stroke [
14‐
16]. There is no clarity of clinical message about any possible contribution of perceived psychosocial stress to stroke risk and a lack of evidence around the potential for stress modification interventions for primary or secondary prevention of stroke [
17]. To date, no systematic review or meta-analysis of studies reporting associations between perceived psychosocial stress and stroke has been published. Psychosocial stressors, caused by relationship, occupational or financial-related stimuli, are recognised as potential contributors to an individual’s perceptions of stress, which is the human response to exposure to psychosocial stressors and inability to cope with the demands made [
18,
19]. However psychosocial stressors are under-investigated compared to more established biological and pathophysiological risk factors for stroke [
17] and the complexity of the relationship between exposure to psychosocial stressors and perception of psychosocial stress is not fully understood. Studies have reported various sub-components of psychosocial stress, including self-perceived stress [
12,
14‐
16] stressful life events (SLE) [
11,
13] and poor adaptation to stress [
10] to be associated with an increased risk of stroke. We performed a systematic review and meta-analysis to evaluate the association between perceived psychosocial stress and stroke, and to clarify differential risks associated with types of stroke and sub-components of perceived stress. Stress as a trigger for stroke events was not included as this was not considered a component of risk. This paper reports the results of the meta-analysis.
Discussion
Meta-analysis of 14 studies (10 cohort, 4 case–control) involving a total of 10,130 strokes found a positive association between perceived psychosocial stress and risk of stroke in adult men and women, suggesting that perceived psychosocial stress may be an independent risk factor for stroke. The combined pooled adjusted effect estimate showed a 33 % increased risk of incident stroke in those reporting perceptions of psychosocial stress and was statistically significant in the separate cohort and case–control study analyses. The increased risk is moderate, being of similar magnitude to risk associated with diabetes mellitus, dietary risk score or depression [
16] when compared to the larger effect size associated with history of hypertension, current smoking, waist-to-hip ratio, alcohol intake, regular physical activity, cardiac causes and ratio of apolipoproteins B to A1. Psychosocial stress is an imprecise term which has multiple interpretations. There is no accepted, universal definition and, depending on perspective, subjective stress comprises physiological, emotional, motivational and cognitive elements all of which may indicate a degree of stress response [
35]. In this review we chose not to narrowly define type of psychosocial stress. An inclusive approach was taken involving wide-ranging descriptions of perceived stress such as general stress, where type of stress was not detailed, occupational stress and major life events [
34]. However the common essential element was the report of subjective stress, as perceived and self-reported by study respondents, not stress that was objectively assessed or measured by another means. The concept of individual perception of stress was important in the absence of any measure or biomarker for actual stress level.
Nevertheless the validity of the exposure measurements of perceived psychosocial stress can be questioned. The majority of the studies included in the meta-analysis used study-specific questionnaires and for many this comprised a single question. The focus of this was broad ranging covering such areas as perceived stress at home or work and including different intensities of exposure from some periods to permanent stress. It is questionable how sensitive the different measures of perceived stress were and indeed their discriminatory properties. A feature to be considered is the extent to which socioeconomic disadvantage was identified, both previous and current, because a sensitive measure of this may explain the observed increase in stroke risk when material disadvantage and associated behavioural hazards are taken into account.
Subgroup analysis to identify specific type of perceived stress associated with stroke was inconclusive with only general stress showing a clear association with increased risk. In nine of the ten cohort studies and two of the four case–control studies general stress indicated perceptions of chronic stress, rather than the result of an individual stressful event or accumulation of stressful life events. These findings potentially resolve some contradictions in the published evidence regarding the type of perceived stress associated with risk of stroke, demonstrating that ongoing perceived stress of a continuous or regular nature was associated with increased risk of stroke. The relative lack of studies measuring discrete life events in this meta-analysis leaves unanswered the question of whether discrete life events contribute to stress-associated stroke risk.
The eligibility criterion of requiring self-report of psychosocial stress limited the studies that could be included in the meta-analysis. This applied to studies of occupational stress in particular where workplace exposure to psychosocial stress was often determined according to a previously developed Job Exposure Matrix of demand and control for a diverse range of occupations rather than self-reported answers to questions about perceived psychosocial stress [
45]. Thus the risks associated with perceived work stress remain speculative.
Subgroup analysis also confirmed that perceived psychosocial stress was linked to stroke in both sexes however higher risk was associated with female sex. There is no clear explanation for this difference and it is not known whether this result indicates that women are exposed to higher levels of stress, whether female perceptions of psychosocial stress are different to male, or whether their reporting experiences of stress are different. Our results concur with those from a literature review of work-stress related stroke among working women, which suggested that work stress may be a more powerful predictor of stroke among women than men [
46]. However female sex specific data are limited and these findings should be investigated further to identify potential explanatory mechanisms. With regard to type of stroke, analysis revealed a significant association between perceived psychosocial stress and fatal stroke of all types but this relationship was not identified in the non-fatal stroke data. Both ischaemic and haemorrhagic stroke were associated with perceived psychosocial stress; however the stronger association was with haemorrhagic stroke. These results contrast with a large individual-participant data meta-analysis of occupational job strain and risk of stroke which found work stress to be associated with a 20 % increased risk of ischaemic stroke but no association with haemorrhagic stroke [
47] and serves to highlight the need for a greater understanding of the biology underlying stress effects.
Potential mechanisms to explain the association by which psychosocial stress may increase risk of stroke are complex and not fully elucidated. Possible explanations relate to impact of perceived psychosocial stress on vascular inflammation, oxidative stress or immune dysfunction underpinning the basic pathophysiology of vascular disease [
36]. Perceived stress is related to increased catecholamine release and sympathetic activation, which may either directly or indirectly affect the vascular system, eg increase thickening of the intima media, progression of carotid arterial disease and impact on blood pressure. Additionally, perceived stress adversely affects immune responses [
37] which may result in increased susceptibility to complications of stroke and thus may contribute to explaining the association between perceived stress and fatal stroke in particular. It may also be the case that individuals with high perceived stress levels have more severe strokes although a mechanism for this has yet to be proposed. Additional to potential pathophysiological mechanisms, studies have also reported adverse behavioural risk profiles with regard to rates of smoking, physical activity and alcohol consumption in those who perceive themselves to be stressed [
38]. The results of the meta-analysis indicating the association between perceived psychosocial stress and stroke, have potential clinical relevance suggesting interventions to manage or reduce perceived stress to be worthy of further investigation, with implications for secondary prevention of stroke. Despite the limited success of interventions to reduce psychosocial stress in primary and secondary prevention of cardiovascular disease [
17], perceived psychosocial stress is theoretically modifiable. It is currently the subject of increased attention through a raised interest in interventions such as mindfulness based stress reduction [
40].
Limitations
The main limitations relate to lack of an agreed definition of perceived psychosocial stress and its measurement, therefore variation and overlap in the perceived psychosocial stress reported may have occurred. As stress was self-reported there is no objectivity in stress measures used, thus understanding and interpretation of what was asked of each participant may vary. A number of studies measured psychosocial stress using a single question, which inevitably will encompass a spectrum of individual interpretations including sensations of anxiety and depression. It is acknowledged that these constructs are difficult to separate out from perceived psychosocial stress and should be considered, although the diagnosed clinical conditions were excluded from the review. Significant heterogeneity was found across the studies which may result from differences in perceived stress measures as well as differences in study design, sample sizes, strategies for analysis and participant characteristics. Several studies examined perceived stress in a younger patient population, where the stroke incidence rate was lower than in those studies with higher proportions of older adults and this may have affected the statistical power to observe significant results. Study selection was limited to English language only, which may have resulted in missing important insights and sample sizes in some studies were small. There was variation in follow-up periods, although none shorter than 6 years and many studies focused on middle-aged men with limited women-specific data, despite the indication that impact of perceived stress may be greater for this group. It was noted that measures of perceived stress in middle age might further be confounded by other previously accumulated adversity relevant to stroke risk. One potential limitation that should be considered is that perceived stress was reported at baseline, with a follow up of 6–30 years and it cannot be assumed that the level of perceived psychosocial stress reported at baseline was consistent or sustained prior to experiencing a stroke.
Competing interests
The authors declare they have no competing interests.
Authors’ contributions
JB led the review process. LC and ML undertook the searches. JB, LC, ML, CC, SJ, CB contributed to study selection and data extraction. JB, LC undertook the meta-analysis. Statistical support was provided by ND. JB, LC wrote the manuscript and all authors read and approved the final manuscript.