Stress Sensitivity and Psychotic Experiences in 39 Low- and Middle-Income Countries

Abstract

Stress has a central role in most theories of psychosis etiology, but the relation between stress and psychosis has rarely been examined in large population-level data sets, particularly in low- and middle-income countries. We used data from 39 countries in the World Health Survey (n = 176 934) to test the hypothesis that stress sensitivity would be associated with psychotic experiences, using logistic regression analyses. Respondents in low-income countries reported higher stress sensitivity (P < .001) and prevalence of psychotic experiences (P < .001), compared to individuals in middle-income countries. Greater stress sensitivity was associated with increased odds for psychotic experiences, even when adjusted for co-occurring anxiety and depressive symptoms: adjusted odds ratio (95% CI) = 1.17 (1.15–1.19) per unit increase in stress sensitivity (range 2–10). This association was consistent and significant across nearly every country studied, and translated into a difference in psychotic experience prevalence ranging from 6.4% among those with the lowest levels of stress sensitivity up to 22.2% among those with the highest levels. These findings highlight the generalizability of the association between psychosis and stress sensitivity in the largest and most globally representative community-level sample to date, and support the targeting of stress sensitivity as a potential component of individual- and population-level interventions for psychosis.

Introduction

Heightened experience of stress is characteristic of psychotic disorders, and theories of psychosis etiology typically include some role for stress in the onset of symptoms.^1–6 Associations between stress and psychosis have been examined and well-replicated using various methods of defining both stress (eg, life events checklists,⁷ ecological momentary assessments,³ stress induction⁸) and psychosis (eg, schizophrenia,⁶ first-episode psychosis,⁹ clinical high risk,¹⁰ psychosis proneness^11,12). However, most studies of psychosis and stress have examined this association in clinical samples, and it remains largely under-studied at the population level. Further, there has been minimal research on this topic outside of wealthy western nations.

There is now substantial epidemiological evidence that psychosis occurs along a continuum within the general population, with many individuals reporting sub-threshold psychotic experiences (PE) (ie, hallucination-like or delusion-like symptoms that do not meet diagnosable criteria due to insufficient intensity, persistence, or associated impairment).¹³ The estimated prevalence of PEs based on meta-analysis is approximately 7.2% of the general population, although world-wide cross-national studies have yielded a broad range (0.7% to 45.8%).¹⁴ Regardless, PEs tend to show similar clinical and functional correlates across these nations (eg, sleep disturbance, pain, and overall health status),^14–16 suggesting that differences in prevalence reflect different “cut-off points” in endorsing PEs between cultures, rather than qualitatively different phenomena.

Prior general population studies have shown that traumatic and stressful life events or exposures can lead to PEs, including a history of abuse,^17,18 bullying,^19,20 sexual assault,²¹ acculturative stress,²² and discrimination,²³ among others. Only 1 prior population-level study, to our knowledge, has specifically examined stress sensitivity as a correlate of PEs. Collip et al¹¹ used experience sampling methodology to assess several domains of stress sensitivity (ie, event stress, activity stress, and social stress) up to 10 times per day for 5 days, finding that greater stress reactivity was associated with an increased likelihood for a trajectory of persistent as opposed to remitting PEs. Although this study has notable strengths in its longitudinal design and method of stress assessment, its sample size (n = 529) was limited relative to typical epidemiological studies, it was not representative of the general population (ie, only included female twins), and, similar to most prior studies of stress and psychosis, it was conducted in a high-income western nation. Therefore, these findings would benefit from additional converging evidence from a globally representative epidemiological sample, particularly with a focus on low- and middle-income countries (LMICs).

Given the high prevalence of poverty in LMICs and the fact that poverty has been linked to stress,²⁴ it is possible that many people in LMICs experience excessive stress. Children who are raised in poverty are more likely to experience multiple stressors²⁵ with a recent study indicating, eg, that poverty may play a role in the occurrence of childhood sexual abuse in LMICs.²⁶ Further, factors such as food insecurity and financial stress may lead to worse mental health in LMICs.²⁷ In short, for many people in LMICs, stressful experiences throughout the life course might increase risk for poorer mental health, including psychosis.

In the current study, we used data from the World Health Survey (WHS) conducted by the World Health Organization (WHO) to determine the role of perceived stress in PE prevalence at the population level, and the consistency or robustness of this finding across a broad range of countries, particularly LMICs where this relation has not been well studied. We hypothesized that stress sensitivity would be closely linked to the prevalence of PEs, corroborating the results of Collip et al’s study,¹¹ as well as other prior epidemiological studies of stress and trauma exposure and PEs.^17–23 Further, given the centrality of stress to most etiological theories of psychosis, particularly its purported role as an intrinsic component of biological vulnerability for psychosis,⁶ we hypothesize that associations between stress sensitivity and PEs will generalize across all included nations.

Methods

The WHS was a cross-sectional survey undertaken in 2002–2004 in 70 countries. Single-stage random sampling was carried out in 10 countries, while stratified multi-stage random cluster sampling was used in the other 60 countries. Survey details are available from the WHO (http://www.who.int/healthinfo/survey/en/). In brief, individuals aged ≥18 years with a valid home address were eligible to participate. All household members had an equal chance of being selected. To ensure comparability across countries, the survey questionnaire was subject to standard translation procedures. Interviews were conducted by trained interviewers with the individual response rate being 98.5% across all countries.¹⁴ Sampling weights were created using the population distribution as reported by the United Nations Statistical Division to adjust for survey nonresponse. Ethical boards at each study site provided ethical approval for the survey with all participants providing informed consent.

Variables

Stress Sensitivity.

Stress sensitivity over the month prior to the interview was assessed by 2 questions: “How often have you felt that you were unable to control the important things in your life?”; and “How often have you found that you could not cope with all the things that you had to do?” The answer options to these questions were: never (score = 1), almost never (score = 2), sometimes (score = 3), fairly often (score = 4), very often (score = 5). The scores of the 2 questions were added to create a scale ranging from 2 to 10. The overall correlation coefficient between these 2 questions was 0.76 in the complete sample of 39 countries. These 2 questions were taken from the Perceived Stress Scale,²⁸ one of the most widely used and validated psychometric measures of stress sensitivity, and have previously been summed and used as a proxy for the perceived stress scale.²⁹ Notably, the Perceived Stress Scale has previously been validated and widely used in many nations, including in LMICs.^30–33

Psychotic Experiences.

Questions on positive PEs were taken from the Composite International Diagnostic Interview (CIDI) 3.0,³⁴ and consisted of dichotomous (yes/no) items related to delusions of control, delusional mood, delusions of reference and persecution, and hallucinations (see supplementary appendix). Those who reported at least 1 of these 4 types of experiences over the past 12-months were considered to have PEs. Previous research has highlighted a high degree of accord between the psychosis module and clinician ratings (delusions: 77.5% agreement; hallucinations: 67.6% agreement).³⁵ Self-reported history of psychotic disorder was assessed with a single item, coded dichotomously: “have you ever been diagnosed to have a mental health problem such as schizophrenia or psychosis?” Analyses were run both with and without respondents with psychotic disorders.

Depression and Anxiety.

The presence of depression in the previous 12 months was established with use of the DSM-IV algorithm based on questions from the World Mental Health Survey version of the CIDI (supplementary material).^36–38 Anxiety was assessed by the question “Overall in the past 30 days, how much of a problem did you have with worry or anxiety?” Respondents could answer: none, mild, moderate, severe, or extreme. Those who answered severe and extreme were categorized as having anxiety.^15,16

Demographics.

Age and sex were reported for all respondents. Age was divided into 3 categories: 18–34 years, 35–59 years, and 60+ years. Sensitivity analyses using age coded as a continuous variable did not meaningfully change any of the results.

Statistical Analysis

From the 69 countries for which there was publically available data, 10 were excluded due to an absence of sampling information. Seventeen of the remaining 59 were also excluded as they either did not collect information on PEs and/or stress sensitivity or >25% of the data on these conditions were missing. Furthermore, Georgia was also excluded due to the poor inter-item correlation of the stress sensitivity measure among this sample (ie, the 2 stress sensitivity items were negatively correlated with each other). Two high-income countries (Spain and the United Arab Emirates) were also excluded as the focus of this study was on LMICs. Details on excluded countries and reasons for their exclusion are provided in the supplementary material (supplementary table S1). Based on the World Bank classification in 2003 (when the survey was conducted), the remaining 39 countries (n = 176 934) consisted of 21 middle-income countries (MICs; n = 83 899) and 18 low-income countries (LICs; n = 93 035). With the exception of China, Comoros, Ivory Coast, India, and Russia, these data are nationally representative.

Statistical analyses were performed with Stata 13.1 (Stata Corp LP). First, the association between stress sensitivity and PEs was examined using multivariable logistic regression analysis with the pooled sample. Two models were constructed: Model 1—adjusted for age, sex, country; and Model 2—adjusted for age, sex, anxiety, depression, country. Country adjustment was undertaken by including dummy variables for each country.¹⁴ Model 2 did not include Morocco as data on anxiety was not collected. Sensitivity analyses that excluded Morocco from Model 1 to enhance comparability did not yield notably different results. Although other factors such as education, wealth, alcohol consumption, and chronic medical conditions (angina, arthritis, asthma, diabetes) were also considered as confounders, since the estimates of stress sensitivity remained largely unchanged even after the inclusion of these covariates, they were not included in the models in order to maximize the sample size (see supplementary appendix for details of these additional covariates and results of regression models including these variables [supplementary table S2]). Second, country-wise multivariable logistic regression models were constructed to assess the association between stress sensitivity (independent variable) and at least 1 PE (dependent variable), adjusting for age and sex. The estimates for each country were also combined into a random-effect meta-analysis.

In the main analysis, we analyzed PEs regardless of whether the respondent had a self-reported history of a psychosis diagnosis, in order to assess the full continuum of mild to severe PEs. In a secondary analysis, we excluded respondents with a psychosis diagnosis to confirm that associations generalized to sub-threshold PEs and were not driven entirely by respondents with psychotic disorders. Taylor linearization methods were used in all analyses to account for the sample weighting and complex study design. Results from the logistic regression analysis are presented as ORs with 95% CIs. The level of statistical significance was P < .05.

Results

Demographic characteristics, stress sensitivity scores, and prevalence of PEs and psychosis diagnosis are shown, by country, in table 1. Prevalence of PEs varied widely by country (ranging from 0.8% in Vietnam to 45.2% in Nepal), as has been shown in previous analyses with these data.¹⁴ PE prevalence was greater in LICs relative to MICs (13.3% vs 9.9%; Chi-squared test P < .001). Stress sensitivity scores likewise varied by country, ranging from 2.85 in Laos to 6.06 in Bangladesh, and were significantly greater in LICs relative to MICs, mean (SE) = 5.09 (0.03) vs 4.54 (0.02), student’s t test P < .001.

Table 1.

Sample Characteristics by Country

Country	N	Age (y)		Female		Stress Sensitivitya		Psychotic Experiencesb		Psychosis Diagnosis
		Mean	SE	%	SE	Mean	SE	%	SE	%	SE
Low-income countries
Bangladesh	5942	36.3	(0.2)	48.5	(0.8)	6.06	(0.05)	13.5	(1.2)	0.66	(0.15)
Burkina Faso	4948	34.6	(0.4)	52.8	(1.4)	5.72	(0.10)	22.5	(1.8)	1.07	(0.20)
Chad	4870	35.8	(0.3)	51.1	(1.2)	5.96	(0.09)	17.1	(1.6)	2.88	(0.33)
Comoros	1836	40.7	(0.5)	50.8	(1.5)	4.86	(0.11)	16.8	(1.8)	0.80	(0.29)
Ethiopia	5089	35.5	(0.3)	51.0	(1.0)	5.48	(0.07)	16.9	(0.8)	1.45	(0.22)
Ghana	4165	36.1	(0.3)	50.9	(1.1)	5.36	(0.06)	5.1	(0.5)	0.69	(0.16)
India	10 687	38.4	(0.3)	47.2	(0.8)	4.87	(0.06)	25.3	(1.1)	2.32	(0.29)
Ivory Coast	3251	34.7	(0.4)	42.2	(1.1)	4.72	(0.07)	22.7	(1.5)	1.12	(0.30)
Kenya	4640	33.4	(0.4)	51.2	(1.5)	5.00	(0.06)	17.0	(1.3)	0.70	(0.20)
Laos	4988	36.8	(0.3)	50.7	(0.9)	2.85	(0.03)	6.0	(0.5)	0.36	(0.11)
Malawi	5551	36.0	(0.4)	51.2	(0.9)	5.19	(0.07)	5.6	(0.6)	1.14	(0.21)
Mauritania	3902	35.8	(0.3)	51.1	(1.5)	5.33	(0.07)	11.5	(1.2)	2.44	(0.49)
Myanmar	6045	38.4	(0.3)	51.1	(0.8)	3.43	(0.08)	2.9	(0.7)	0.33	(0.12)
Nepal	8820	37.1	(0.2)	49.5	(0.7)	4.65	(0.04)	45.2	(0.8)	2.59	(0.20)
Pakistan	6501	36.6	(0.3)	49.6	(0.9)	5.45	(0.05)	2.3	(0.3)	1.11	(0.20)
Senegal	3461	35.3	(0.3)	50.9	(1.3)	4.83	(0.07)	18.8	(1.3)	1.35	(0.34)
Vietnam	4174	38.4	(0.5)	51.3	(1.2)	2.94	(0.10)	0.8	(0.2)	0.07	(0.03)
Zambia	4165	35.2	(0.3)	51.1	(1.1)	4.76	(0.06)	10.4	(0.7)	0.72	(0.17)
Middle-income countries
Bosnia Herzegovina	1031	44.0	(0.6)	51.1	(2.2)	4.14	(0.12)	1.9	(0.5)	0.09	(0.06)
China	3994	45.1	(0.7)	51.0	(1.6)	3.86	(0.10)	5.7	(1.7)	0.34	(0.10)
Croatia	993	49.5	(0.7)	58.1	(1.8)	4.22	(0.08)	9.0	(1.0)	2.01	(0.51)
Czech Republic	949	45.8	(0.9)	52.1	(2.1)	4.12	(0.09)	9.2	(1.4)	0.50	(0.24)
Dominican Republic	5027	38.5	(0.4)	49.1	(1.2)	4.22	(0.05)	22.0	(1.3)	0.89	(0.20)
Ecuador	5675	38.3	(0.4)	48.9	(1.2)	4.12	(0.08)	9.3	(1.0)	0.96	(0.23)
Estonia	1020	47.1	(0.5)	55.4	(2.0)	4.06	(0.07)	12.3	(1.1)	1.39	(0.43)
Kazakhstan	4499	41.4	(1.1)	52.1	(1.7)	4.18	(0.11)	3.3	(0.6)	0.49	(0.13)
Latvia	929	46.5	(0.9)	55.4	(2.3)	5.45	(0.10)	13.4	(1.6)	0.66	(0.33)
Malaysia	6145	38.8	(0.3)	49.6	(0.8)	3.11	(0.03)	7.6	(0.5)	0.21	(0.06)
Mauritius	3968	40.6	(0.4)	50.8	(1.0)	4.48	(0.09)	8.2	(1.1)	0.61	(0.15)
Morocco	5000	37.7	(0.5)	50.5	(1.4)	4.89	(0.09)	18.7	(1.3)	0.67	(0.24)
Namibia	4379	37.0	(0.4)	53.0	(1.2)	4.94	(0.07)	11.7	(0.8)	2.95	(0.41)
Paraguay	5288	37.1	(0.3)	50.4	(0.9)	4.16	(0.03)	9.6	(0.6)	0.48	(0.11)
Philippines	10 083	37.2	(0.2)	50.4	(0.7)	5.09	(0.05)	8.8	(0.7)	0.38	(0.07)
Russia	4427	51.4	(0.7)	64.4	(1.6)	4.91	(0.08)	8.8	(1.2)	0.40	(0.10)
South Africa	2629	37.5	(0.5)	52.0	(1.6)	4.39	(0.09)	15.5	(1.6)	1.18	(0.28)
Sri Lanka	6805	40.5	(0.4)	47.9	(0.7)	3.92	(0.05)	2.8	(0.4)	0.65	(0.17)
Tunisia	5202	38.6	(0.3)	50.1	(0.9)	5.00	(0.06)	15.5	(1.2)	1.86	(0.25)
Ukraine	2860	46.1	(0.6)	54.5	(1.5)	4.70	(0.08)	7.5	(0.9)	0.65	(0.17)
Uruguay	2996	45.0	(0.6)	52.5	(3.0)	3.92	(0.08)	5.6	(1.1)	0.72	(0.11)

Note: SE, standard error. Data are based on weighted data. Income level was based on the World Bank classification in 2003.

^aThe stress sensitivity scale ranged from 2 to 10 with higher scores indicating greater levels of stress sensitivity.

^bAt least 1 of the 4 following psychotic experiences: delusional mood, delusions of reference and persecution, delusions of control, and hallucinations.

Stress sensitivity was associated with increased odds of reporting PEs in logistic regression analyses, OR (95% CI) = 1.22 (1.20–1.25), which was slightly attenuated but still significant when adjusted for anxiety and depression, OR (95% CI) = 1.17 (1.15–1.19) (table 2). The strength and significance of the association between stress sensitivity and PEs were unchanged when excluding individuals with a self-reported diagnosis of a psychotic disorder. Furthermore, results were nearly identical when using hallucination-like experiences only or delusion-like experiences only as the outcome variable (supplementary tables S3 and S4). Notably, the ORs represented the increased risk for PEs associated with each additional point on the stress sensitivity scale (range 2–10). Therefore, while each point on the stress sensitivity scale yielded an incremental increase in odds for PEs, this translated into drastic differences in prevalence of PEs between those scoring at the lower range of the scale (prevalence = 6.4% for respondents with the minimum score) and those at the upper range (prevalence = 22.2% for respondents with the maximum score), as illustrated (without adjustments) in figure 1.

Table 2.

The Association Between Stress Sensitivity (Independent Variable) and At Least 1 Psychotic Experience (Dependent Variable) Estimated by Logistic Regression

					Individuals With Psychosis Diagnosis Excluded
	Model 1		Model 2		Model 1		Model 2
Characteristic	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Stress sensitivitya	1.22***	[1.20,1.25]	1.17***	[1.15,1.19]	1.22***	[1.19,1.24]	1.16***	[1.14,1.19]
Covariates
Sex
Male	1.00	[1.00,1.00]	1.00	[1.00,1.00]	1.00	[1.00,1.00]	1.00	[1.00,1.00]
Female	1.06	[0.97,1.16]	0.99	[0.90,1.09]	1.07	[0.97,1.17]	1.00	[0.91,1.11]
Age (y)
18–34	1.00		1.00		1.00		1.00
35–59	1.02	[0.96,1.08]	0.95	[0.89,1.02]	1.01	[0.95,1.08]	0.95	[0.89,1.02]
≥60	0.93	[0.83,1.05]	0.79***	[0.70,0.90]	0.94	[0.83,1.06]	0.80***	[0.70,0.91]
Anxiety			2.36***	[2.12,2.62]			2.28***	[2.05,2.54]
Depression			2.73***	[2.37,3.14]			2.70***	[2.33,3.14]

Note: Model 1—Adjusted for sex, age, and country. Model 2—Adjusted for sex, age, anxiety, depression, and country. Model 2 does not include Morocco as information on anxiety was not collected.

^aThe stress sensitivity scale ranged from 2 to 10 with higher scores indicating greater levels of stress sensitivity.

***P < .001.

Fig. 1.

Prevalence (%) of at least 1 psychotic experience by stress sensitivity scale score. Estimates are based on weighted data, using the pooled analysis of all 39 countries. Individuals with a psychosis diagnosis are included in the analysis. Bars denote 95% CIs. The stress sensitivity scale ranged from 2 to 10 with higher scores indicating greater levels of stress sensitivity.

In a second set of analyses, logistic regression analyses were conducted within each individual country and then meta-analyzed in order to (1) examine the generalizability of findings across the constituent nations of this data set; (2) provide a visualization of variability in strength of associations; and (3) to confirm the robustness of findings using a second statistical approach. ORs indicated that the association between stress sensitivity and PEs was greater than 1 for all examined countries, and that there were statistically significant associations in 37 out of the 39 countries (figure 2). Notably, the meta-analysis yielded an OR of 1.23, confirming the robustness of the pooled logistic regression estimate (OR = 1.22). As in the logistic regression analyses using the overall sample, results were generally unchanged when excluding individuals with a self-reported diagnosis of a psychotic disorder (supplementary figure).

Fig. 2.

The country-wise association between stress sensitivity (independent variable) and at least 1 psychotic experience (dependent variable) estimated by logistic regression adjusted for age and sex. The stress sensitivity scale ranged from 2 to 10 with higher scores indicating greater levels of stress sensitivity. Individuals with a psychosis diagnosis are included in the analysis.

Discussion

The main finding of this study is that stress sensitivity is associated with increased odds for PEs, regardless of the presence of a psychotic disorder diagnosis. While the relation between stress sensitivity and psychosis has long been integral to theories of schizophrenia etiology, and has been previously shown in many clinical samples, this is incomparably the largest and most globally representative sample in which this association has been demonstrated to our knowledge. Further, our within-country analyses demonstrate that this association is robust across nearly all studied cultures.

These findings concur with prior population-based studies, which have likewise demonstrated increased odds for PEs among respondents reporting exposure to variably defined traumatic or stressful events or phenomena, such as childhood abuse^17,18 or acculturative stress.²² In particular, they provide converging evidence with the finding that heightened reactivity to daily stressors, measured in real-time using experience sampling methods, is associated with more severe or persistent PEs.¹¹ This may be particularly relevant in the LMICs given the low rates of mental health service utilization in many of these nations³⁹ and consequently elevated risk for the progression of mental illnesses towards greater severity and persistence.⁴⁰ Further, mean stress sensitivity levels and PE prevalence were greater in LICs vs MICs in these data, suggesting that individuals may face higher levels of daily stress (and consequently greater prevalence of PE) in nations with lower average income.

Societal and Country-Specific Mechanisms

Although a relationship was observed between stress sensitivity and PEs across all countries, 11 of the 15 strongest effect sizes were observed in MICs, which may relate to both country-wise processes and country-specific factors. In terms of the former, eg, by 2008 although the incidence of poverty was higher in LICs, most of the world’s poor were actually living in MICs,⁴¹ which may be important for the stress-PEs association observed in MICs. Alternatively, other societal processes might have also been important. The collapse of the Soviet Union and subsequent transition to a market economy generated high levels of societal stress⁴² and comparatively worse mental health,^43,44 but low levels of help-seeking behavior due to the social stigma attached to mental illness⁴⁴ in conditions of inadequate service provision.^44,45 This might account for the strong association in former Soviet MICs such as Estonia, Latvia, and Ukraine. Other factors such as rapid urbanization that has occurred in countries such as Malaysia and China since 1990⁴⁶ which has been linked to stress and worse mental health in migrants,⁴⁷ might also be important in this context, especially as urbanicity has itself been associated with increased risk for psychosis.⁴⁸ Country-specific factors might include the experience of war. It is notable that the country with the highest OR was Bosnia Herzegovina, where there was a large-scale war in 1992 to 1995 that resulted in nearly 100 000 deaths and damage to 60% of houses.⁴⁹ Direct experience of war may be particularly relevant, as an earlier study showed that women with war exposure were not only more likely to develop post-traumatic stress disorder (PTSD), but also had more pronounced psychological symptoms such as paranoid ideation and psychoticism compared to women in a control group indirectly exposed to war, with exposure to a greater number of traumatic events linked to more prominent symptoms. In addition, experiencing a higher number of war stressors was also associated with having more post-war stressors, which in turn, were linked to psychological symptoms independent of traumatic war experiences.⁵⁰

Biological Mechanisms

Stress sensitivity has been implicated in the diathesis stress model^6,51 and other prominent models as a causal factor in both the onset and exacerbation of psychotic symptoms. The underlying biological mechanism may be elevated baseline activity and responsivity of the hypothalamic-pituitary-adrenal (HPA) axis stress response system,⁵² in combination with reduced inhibitory feedback to the HPA-axis, potentially due to reduced expression of glucocorticoid receptors among people with psychotic symptoms.⁵³ Interactions between the HPA-axis and dopaminergic systems may lead to elevated dopamine levels in some brain regions, which can be expressed behaviorally as PEs,⁶ even in the context of mild to moderate stressors.³ Biomarker studies have typically found support for these mechanisms even among individuals with sub-threshold PEs, including elevated basal cortisol secretion,^54–56 reduced connectivity between the striatum and frontal cortex,⁵⁷ and increased dopamine reactivity (ie, elevations of plasma hypovanillic acid following metabolic perturbation) in response to daily life stress.⁵⁸ Neuroimaging studies of individuals with schizophrenia or at clinical high risk for psychotic disorder have found increased metabolism in the hippocampus, which is a site of glucorticoid receptors and therefore relevant to stress sensitivity,⁵⁹ and evidence for increased dopamine release in the striatum in the context of induced stress.⁸ Such biological mechanisms would necessarily be universal despite what are likely broad differences in stress exposure across nations, and therefore are consistent with our findings that stress sensitivity is similarly related to odds for PEs regardless of nation or culture. Notably, the association between stress and psychosis was not explained by co-occurring anxiety or depressive symptoms, which were themselves strongly associated with PE. This is in concurrence with past work on youth at clinical high risk for psychotic disorder, among whom salivary cortisol appears to be independently related to anxiety and psychotic symptoms.⁵⁴

Clinical and Public Health Implications

The primary clinical implication of an association between stress sensitivity and psychosis is that stress alleviation techniques may be beneficial for people with PEs. Interventions that alleviate perceived stress may be efficacious in reducing symptom severity among individuals with psychotic disorders, or may be potentially efficacious in preventing or delaying psychotic disorder onset among help-seeking youth with sub-threshold psychosis (ie, youth at clinical high-risk for psychotic disorder).⁵¹ Various efforts to explore this possibility have yielded positive results. For example, mindfulness training has been shown to have moderate efficacy for people with psychotic disorders in a recent meta-analysis.⁶⁰ Physical exercise, which can reduce stress,⁶¹ has likewise been shown to alleviate psychotic symptoms in another recent meta-analysis.⁶² Stress sensitivity can be directly addressed through psychotherapeutic approaches such as cognitive-behavioral therapy, which has shown to be efficacious in treating psychotic symptoms,^63,64 even in individuals with schizophrenia who are not taking anti-psychotic medications.^65,66 Stress alleviation is likewise a key component of attempts to prevent or delay onset of schizophrenia in help-seeking youth at clinical high-risk,⁶⁷ among whom cognitive therapy and supportive therapy appear to be beneficial.^63,68

It is unclear whether such individual-level approaches would benefit individuals in the general population with PEs, given that many of these individuals may not be help-seeking (although they are more likely to be in treatment than individuals without PEs).^69,70 Public mental health strategies may therefore be more appropriate, given the broad range of mental health outcomes associated with stress exposure in conjunction with the high rates of stress exposure in LMICs. For example, reducing stress sensitivity may become a key component of clinical staging approaches to prevention, which target broader phenotypes that include aspects of psychosis, depression, anxiety, and mania, without regard for specific diagnostic outcomes.⁷¹ Personal control of sensitivity to stressors is largely influenced by environmental factors, and varies across countries at the aggregate level.⁷² Thus, macro-level efforts can focus on alleviating structural stressors that are prominent in LMICs, such as inequities in educational attainment,⁷³ meaningful employment,⁷⁴ and neighborhood safety.⁷⁵ Such strategies may lead to a reduction in PEs (assuming that the relation between stress and psychosis is, to some extent, causal), as well as a range of other mental health outcomes, even without directly targeting psychosis. Finally, some countries in our study experienced major social change or upheaval shortly before or even during survey administration (eg, war in Bosnia Herzegovina, or the collapse of the Soviet Union, which affected several included nations), highlighting the potential value of public mental health efforts during periods of national crisis.

Strengths, Limitations, and Conclusions

Strengths of the present study include the use of a very large multinational data set, focused on LMICs that are under-represented in the prior research literature on this topic. The use of data from 39 countries essentially allowed a series of replications of the primary finding within a single study, and confirmation of the strength of the association using meta-analysis. The primary limitation to this study is the use of cross-sectional data, which makes it difficult to infer the direction of causality between heightened stress sensitivity and PEs. This is particularly relevant given that the stress measures focused on a 30-day time-period, whereas PEs were assessed in reference to the 12-month period prior to data collection. Similarly, the time-frame also varied between the depression (12-month) and anxiety (30-days) measures. In addition, we used an abridged version of the perceived stress scale, which potentially differs in validity and reliability compared to the original complete scale, and may likewise vary psychometrically across cultures.

The association between stress sensitivity and psychosis has long been supported in clinical samples, but has rarely been examined in general population samples, particularly in LMICs. These data support the generalizability of this relation across a global general population sample, further supporting theories that include stress sensitivity as a key component of psychosis etiology and phenomenology. Clinical and public health approaches to psychosis treatment and prevention may benefit from a focus on stress sensitivity at the individual level, and building resilience and reducing exposure to adversity at the population level.

Supplementary Material

Supplementary material is available at http://schizophreniabulletin.oxfordjournals.org.

Funding

A.K.’s work was supported by the Miguel Servet contract financed by the CP13 / 00150 project, integrated into the National R + D + I and funded by the ISCIII - General Branch Evaluation and Promotion of Health Research—and the European Regional Development Fund (ERDF-FEDER).