Psychological Stress and Gut Microbiota Composition: A Systematic Review of Human Studies Open Access

The bi-directional dialog between the gut-brain axis and the role of psychological stress has recently become the focus of intense debate. Given that many trillions of bacteria reside and have co-evolved in the human gastrointestinal (GI) tract, this has resulted in a symbiotic relationship where the GI tract offers a hospitable environment for these bacteria, which in turn perform a plethora of roles implicated in the support of human health above and beyond the fermentation of non-digestible substrates [1]. Perhaps, one of the most important of these is the modulation of the central nervous system, with mounting evidence demonstrating that psychological stress can mediate the diverse range of microorganisms residing in the human intestine [2]. However, despite this, the relationship remains poorly understood.

Psychological stressors can come in many forms, ranging from those experienced during early childhood, through adulthood, and have been shown to impact upon the composition of the gut microbiota. For example, acute stress can increase the presence of catecholamines, which can increase levels of pathogenic bacteria by up to 10,000 fold in vitro, as well as causing localized shifts in microbiota habitats [3]. Chronic stress has also been implicated in long-term modifications of the gut microbiota in animal models that can result in differences in alpha diversity, which appear to differ dependent on the characteristics of the stressors [4]. Examples are that the gut microbiota of students suffering from academic stress is composed of fewer health-promoting bacteria [5], and very recent evidence has revealed dysbiosis in the microbiota of frontline healthcare workers who experienced psychological stress while working during the COVID-19 pandemic [3, 6]. Therefore, both acute and chronic stress can cause morphological and functional changes in the human gut microbiota.

Despite these associations of acute and chronic psychological stress with gut microbiota composition, the underlying mechanisms remain elusive, with evidence suggesting that the dysregulation of tryptophan metabolism may be a contributing factor [7]. Regardless of the implicated mechanisms, it is important to note that when stressors are removed in rats, the relative microbial abundance in terms of both species and distribution appear to resolve [8]. However, interestingly, the microbial metabolic profiles remain altered, potentially suggesting that the recovery time for gut microbial metabolic functionality from stressful experiences may be slow [8].

This impaired recovery time may have important implications in terms of aging, where the composition of the gut microbiota and their metabolic outputs in the blood of older adults has been shown to deteriorate with declining health and has been proposed as a modifiable predictor of healthy trajectory [9]. In addition to this, disparities in the gut microbiota have also been observed between genders as well as across ethnicities [10, 11]. For example, a cohort study showed that age and gender were among the 69 factors which were shown to correlate significantly with overall microbiome community variation [12]. Substantial ethnic differences in the composition of the gut microbiota have also been found [13]. Although this is likely to be influenced by a plethora of extrinsic factors, psychological stress is known to be vitally important [10, 14]. Furthermore, the type of stress may be a modifier of the associations between stress and gut microbiota [15]. However, further work is required to fully understand its role in dysbiosis across different demographics.

In summary, the ability of acute and chronic psychological stress to modulate the gut microbiota composition is yet to be fully elucidated. Despite this, evidence exists which suggests that the role of psychological stress on the gut microbiota can differ across age, ethnicity, and the type of psychological stress and between gender [16]. Our systematic review will, for the first time, synthesize this literature and provide a summary of the findings on how age, gender, ethnicity, and the types of stress may moderate the associations between psychological stress and gut microbiota composition, offering insights based on the totality of the evidence.

This study was developed and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [17] and other standards (e.g., Johnson and Hennessy [18], 2019). The review protocol is listed in the International Prospective Register of Systematic Reviews (PROSPERO) (CRD: 42021284124).

A literature search was performed in electronic bibliographic databases: PubMed, Web of Science, PsycINFO, and Embase, from inception to November 2021. Search strategies included combinations of terms related to acute and chronic psychological stress, gut microbiota, and human study. A summary of search strategies was provided in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000533131). Furthermore, to minimize the effect of publication bias, a snowball method, characterized by manual checking of references from retrieved studies, was applied to relevant studies that met the selection criteria outlined below.

Search results were merged into the reference management software EndNote. Two authors (Y.T.Y. and Y.L.) independently reviewed the records, in case of disagreement between the two authors, a third author (M.L.) was consulted to make the final decision.

Studies were included if they (a) were cross-sectional studies, case-control studies, longitudinal studies, randomized controlled trials, clinical trials with either parallel, crossover, or cohort design; (b) examined the associations between psychological stress and gut microbiota, or the associations between gut microbiota and psychological stress in humans; (c) psychological stress included at least acute laboratory stress, chronic stress, stressful life events, perceived stress, or early life stress; (d) had samples that did not overlap with other identified studies (if data or data subsets were from duplicate publications, only publications with the largest sample size was included); and (e) were published as full-text articles in peer-reviewed scientific journals. Studies were excluded if they were (a) without human subjects or (b) were qualitative studies, case reports, editorials, protocols, meta-analyses, or reviews.

In this study, assessments of the biodiversity and composition of microbiota were based on stool sample testing using culture-dependent methods, real-time polymerase chain reaction, fluorescence in situ hybridization, or pyrosequencing for bacterial 16s ribosomal ribonucleic acid genes. The alpha diversity and beta diversity were used to assess the biodiversity of gut microbiota. The alpha diversity provides a summary statistic of the microbial community, whereby a higher alpha diversity indicates a greater number of species, with more even representation, and/or greater biodiversity according to the ancestral dissimilarity of species. Beta diversity is an inter-individual measure that examines similarity of communities relative to the other samples analyzed. All bacterial information was reviewed and selected before the final analyses, including bacterial taxonomy, percentage, and relative abundance at different taxonomic levels. For consistency, the included studies were analyzed at the phylum-, family-, and genus-levels in the current study. We contacted the investigators of the eligible studies if we were unable to extract data on bacterial abundance from the published studies.

Data were extracted by two independent authors, and disagreements were reviewed and resolved by the first author. Data extracted included the following: (a) study level (country, author, publication year, study design, type and measurement of stress, microbiota diversity assessment method and estimation, confounding variables), (b) sample level (participants’ age, sex, and ethnicity), and (c) effect sizes (analyses which compared the gut microbiota between cases and controls, or assessed associations between psychological stress, and the gut microbiota).

Pearson correlation coefficient r, β coefficient, odds ratio, or standardized mean difference was used to determine the associations. They were set to be positive (greater than the null value) for associations that favored the hypotheses. When a study reported both overall effect size and subgroup effect sizes, such as by sex, the overall effect size was used in the overall analysis.

Two authors independently assessed the quality of the studies using the U.S. National Institutes of Health Study Quality Assessment Tools (National Heart Blood and Lung Institute, 2019). The assessment tool rates each study based on 14 criteria. For each criterion, a score of one was assigned if “yes” was the response, whereas a score of zero was assigned otherwise (i.e., an answer of “no,” “not applicable,” “not reported,” or “cannot determine”). Overall quality was rated based on the total score of the scale: “7≤ total score” = good, “4< total score <7” = fair, “total score <4” = poor. The risk of bias of each study decreased with the increase in the total score (online suppl. Table 2).

The literature search identified 13 eligible studies that investigated the associations between psychological stress and the composition of the gut microbiota as a primary outcome. The 13 studies were summarized in Table 1.

The study selection process that followed the literature search is summarized in Figure 1. A total of 10,790 citations were identified through four databases. After initial screening and removal of duplicates, 10,398 records remained, of which 10,205 were screened by abstract and excluded. Following the initial screening, 193 full-text studies were retrieved for detailed review and were assessed against the established selection criteria.

One hundred and eighty full-text records did not fulfill the set inclusion criteria and after their removal, 13 studies were included in the current review. All the included 13 studies had good quality according to the National Heart, Lung, and Blood Institute’s Quality Assessment Tool (online suppl. Table 2).

The characteristics of the included studies are summarized in Table 1. The studies were conducted in 7 countries including the USA (5 studies) [19, 22, 26, 27, 29], China (2 studies) [20, 28], South Africa (2 studies) [21, 30], and one study each from Switzerland [23], Belgium [24], the Netherlands [25], and Australia [5]. Most of the studies were of cross-sectional (n = 4) [19, 24, 25, 29] and case-control design (n = 5) [5, 26‒28, 30], which could not determine potential causal relationships between psychological stress and microbiota composition. The medium sample size of the studies was 70, ranging from 22 to 204. The medium age of participants was 28.0 years, ranging from 26.5 days to 82 years. Three studies targeted at children [20, 24, 27], and the remain ten studies targeted at adults [5, 19, 21‒23, 25, 26, 28‒30]. About 39.0% of the studies (n = 5) were conducted among participants with medical conditions such as Crohn’s disease, inflammatory bowel diseases, major depressive disorder, and posttraumatic stress disorder and trauma-exposed. Except for three studies involving Caucasian [22], Hispanic [26], and South African mixed ancestry ethnic groups [30], the other studies did not report the ethnicity of participants.

Eleven of the 13 studies assessed alpha diversity [19‒26, 28‒30] and nine assessed beta diversity [19, 21‒26, 28, 30] of the gut microbiome. The most frequently used measurement method was 16S ribosomal ribonucleic acid gene sequencing. Four indices were used to assess alpha diversity, including estimates of richness (observed species and Chao 1), richness/evenness (Shannon and Simpson indices). Four indices were used to assess beta diversity, including Aitchison distance metric, Bray-Curtis dissimilarity matrix, UniFrac (unweighted or weighted), and computed based on the “w” metric. Shannon and Simpson indices were the most frequently used indices to assess alpha diversity, and UniFrac was the most frequently used index to assess beta diversity. Perceived stress was the main type of stress (53.85%, n = 7) [5, 19, 20, 22, 23, 28, 29], and perceived stress was mainly measured by the Perceived Stress Scale-10 (30.77%, n = 4) in the included studies [19, 22, 28, 29].

Summaries of the associations of alpha diversity and beta diversity of gut microbiota with psychological stress are shown in Table 2; Figure 2, and online supplementary Figure 1. Of the 11 studies reporting the alpha diversity and beta diversity [19‒26, 28‒30], more than half of them did not report associations of alpha diversity and beta diversity of gut microbiota composition with psychological stress [19‒26, 28‒30]. Among the studies reporting such associations, two studies reported psychological stress to be associated with smaller alpha diversity measures of Shannon index, Chao 1, and Simpson index [23, 24]; no significant associations were observed between observed species and psychological stress. Only one study reported that beta diversity assessed by UniFrac (weighted) was negatively associated with psychological stress [24]. However, other studies found that psychological stress was not significantly associated with indicators of beta diversity [19, 21, 22, 25, 26, 30]. When stratified by the type of psychological stress, only perceived stress and self-reported stressful life events were significantly associated with alpha diversity and beta diversity of gut microbiota. One study showed that perceived stress was associated with the lower Shannon index, Chao 1, and Simpson index [20, 23]. One study reported that self-reported stressful life events were associated with the lower Simpson index [24]. Six of the 13 studies controlled for confounding variables to examine the associations of psychological stress with the alpha diversity and beta diversity of gut microbiota.

All 13 included studies assessed associations between psychological stress and the relative abundance of gut microbial taxa at phylum-, family-, and genus-levels (Table 2; Figure 3; online suppl. Fig. 2).

• 1.

  At the phylum-level, relative abundance of seven phyla were identified associated with psychological stress, including: Actinobacteria, Bacteroidetes, Euryarchaeota, Firmicutes, Proteobacteria, Tenericutes, and Verrucomicrobia. The following results were found in at least two studies. Psychological stress was consistently associated with increased abundance of Euryarchaeota in two studies [24, 25], and decreased abundance of Proteobacteria [20, 23, 28] and Verrucomicrobia in three studies [24, 25, 30]. Findings for psychological stress and other microbiota at phylum-level were not consistent.

• 2.

  At the family-level, relative abundance of 17 taxa was identified in 13 studies that were significantly associated with psychological stress. Only Alcaligenaceae was found to be consistently and negatively associated with psychological stress from two studies [23, 28]. The results for other microbiota at family level were only reported in individual studies, and most of these found that microbiota at family-level, such as Erysipelotrichaceae and Veillonellaceae were negatively associated with psychological stress.

• 3.

  At the genus level, relative abundance of 41 genera was found to be associated with psychological stress. The following results were found in at least two studies: the relative abundance of Lachnospira [22, 23], Lachnospiraceae NK4A136 [24, 25], Phascolarctobacterium [24‒26], Sutterella [23, 28], and Veillonella [24, 27] were decreased while Methanobrevibacter [24, 25], Rhodococcus [24, 25], and Roseburia [24, 25] were increased in the higher psychological stress group compared to the lower stress group.

When stratified by the type of psychological stress, at the phylum-level, maternal prenatal stress and negative emotions were consistently associated with the increased abundance of Euryarchaeota [25]; early life stress and maternal prenatal stress were consistently associated with the decreased abundance of Verrucomicrobia [25, 30]. At the genus-level, self-report stressful life events and negative emotions were consistently associated with the reduced abundance of Bacteroides [24]; maternal prenatal stress and negative emotions were associated with the reduced abundance of Methanobrevibacter, Rhodococcus, and Roseburia [24, 25]. Maternal prenatal stress and negative emotions were associated with the increased abundance of Lachnospiraceae NK4A136 [24, 25]; early life stress, maternal prenatal stress, and negative emotions were associated with the increased abundance of Phascolarctobacterium [24‒26]; daily stress experiences of infants during care in the NICU and negative emotions were associated with the increased abundance of Veillonella [24, 27].

This study synthesized the literature to determine associations between psychological stress and the gut microbiota from human studies. Our findings show, based on data from limited studies targeted at children and adults, that there is an overall negative association between psychological stress and alpha diversity measures including Shannon, Chao 1, and Simpson index. Most of the studies reported that psychological stress was not significantly associated with indicators of beta diversity. We also reveal more nuanced changes which occurred in the composition of the microbiota at phylum-, family-, and genus-levels, and such changes were modified by the type of psychological stress. We highlighted the challenges in determining the impact of psychological stress on the gut microbiota across gender, age, and ethnicity given the current available literature.

Although previous studies have demonstrated a negative association between psychological stress and the alpha diversity of the gut microbiota in children [24] and adults [25], ours is the first to show this using aggregated data. Furthermore, it should also be noted that whereas more than half of the studies did not report such relationships, those who found significant associations reported negative ones. These declines in microbial integrity occurring in tandem with increased psychological stress are also in line with previous research conducted in both animal models [31] and human participants [3, 6]. Although the exact mechanisms which underpin this remain unclear, it is known that the gut microbiota can respond to mammalian hormones and neurotransmitters, as well as other aspects of GI physiology, which may result in a microbial milieu characterized by a maladaptive immune response to stressors, and ultimately lead to a compromised ability to survive in an ever-changing environment [31].

Moreover, we found that there were nuanced changes to the phylum-, family-, and genus- of the microbiota. Most studies which reported results were concerned with the decreased abundance of Proteobacteria and Verrucomicrobia, and increased abundance of Euryarchaeota. Proteobacteria, one of the most abundant phyla in the human gut microbiota, was shown to be negatively associated with psychological stress in children and adults [32]. This is an unexpected finding since expansions in Proteobacteria have been previously observed in mouse models of chronic stress [33]. Especially when considering that previous studies have also demonstrated that a proliferation of pro-inflammatory Proteobacteria can disrupt gut-brain communication, causing activation of microglia cells as well as contributing to increased adiposity [34]; both of which have been associated with psychological stress [35, 36]. However, previous results were based on animal models, whereas our findings were summarized from human studies, therefore the findings were inconsistent. More studies are needed to clarify the associations between psychological stress and the abundance of Proteobacteria.

Despite this, some of our findings were better aligned with the literature. For example, we determined that Verrucomicrobia was negatively related to psychological stress. Verrucomicrobia is a phylum of gram-negative bacteria and encompasses genera such as Akkermansia, which is a mucus-degrading bacterium and has been shown to infer beneficial properties in humans [30, 37], such as anti-inflammatory effects and the ability to induce regulatory T cells in adults [30]. From the studies included in the current review, Michels et al. [24] also found that the abundance of Akkermansia was negatively associated with chronic stress [38], which supports the findings on Verrucomicrobia phylum. Furthermore, it also supports the hypothesis that posits that a decreased exposure to anti-inflammatory bacteria can lead to an increased vulnerability to stress in mice [38]. The negative relationship between Verrucomicrobia and psychological stress as demonstrated by our findings is therefore in agreement with the wider literature and provides further evidence for a bi-directional relationship which governs the gut-brain axis [39].

In addition to this, we also found an increase in the abundance of Euryarchacota in those experiencing high levels of psychological stress. Euryarchacota is a phylum of archaea which consist of only a handful of organisms, and make up a small proportion of the gut microbiome [40]. Despite this Euyarchacota has been shown to infer pathogenic effects which are possibly thought to result from their ability to facilitate interspecies hydrogen transfer [41]. The interspecies hydrogen transfer in humans might support the growth of fermenting bacteria, which themselves could be either true pathogens or at least opportunistic pathogens [41]. The second mechanism of the negative impacts of Euyarchacota on health may be via the transformation of heavy metals to toxic derivatives, which are known to be more toxic than the original compounds. Our findings highlight how Euryarchacota is positively associated with psychological stress and given the pathological nature of these organisms this is perhaps a cause for concern since these archaea may be a contributing factor toward some of the poor health outcomes often attributed to psychological stress in adults [42].

At genus-level, we found that psychological stress was associated with change in the relative abundance of specific gut microbiota. For example, psychological stress was associated with a lower abundance of Lachnospiraceae NK4A136 in two studies [24, 25]. Some studies also reported that other mental health problems, such as depression in adults [43], were associated with a decreased abundance of Lachnospiraceae NK4A136, which is a butyrate-producing bacterium. Therefore, our findings provide further evidence that mental health problems may have detrimental effects on the abundance of Lachnospiraceae in children [44]. In addition to this, the abundance of Roseburia was found to be positively associated with psychological stress in the current study. This is an unexpected finding since Roseburia also belongs to the family of Lachnospiraceae. The associations of psychological stress with Lachnospiraceae NK4A136 and Roseburia were opposite in the current study. This may be related to the different measures of psychological stress. In the included studies, psychological stress measures of negative emotions and low happiness were positively associated with the abundance of Roseburia. However, psychological stress measures of happiness and parasympathetic activity were negatively associated with the abundance of Lachnospiraceae NK4A136. Further, some studies have indicated that Roseburia is an important butyrate-producing bacterium, which has beneficial effects on energy metabolism and intestinal barrier integrity in adults [45]. Therefore, more studies are warranted to determine the associations between different measures of psychological stress and the genera of Lachnospiraceae NK4A136 and Roseburia.

As well as aiming to determine the association between psychological stress and the gut microbiota, we also aimed to investigate the influence of age, gender, and ethnicity on such associations. Our findings show that existing studies generally recruited humans from a diverse range of ages, spanning from 26.5 days to 82 years. This is particularly important since previous research illustrates how the diversity of the gut microbiota declines in later life [9]. That said, many studies did not stratify by age and therefore it is challenging to determine whether the negative relationship between psychological stress and the microbiota is exacerbated over an individual’s lifespan and is an area which warrants further future research. Similarly, although many of the available studies were well-represented by both males and females, there was a lack of stratification making it difficult to determine the impact gender may have upon the relationship between psychological stress and the gut microbiota. As such, this is an area of research which also warrants future attention, especially since previous evidence suggests that gender may indeed have an impact [10]. In terms of ethnicity, many of the studies used in our analysis did not report this and those which did investigated predominantly Caucasian or Hispanic populations [11]. Therefore, further future research to determine the impact of ethnicity upon the relationship between stress and the gut microbiota is required in order to supports or refute existing evidence which suggests a link.

The primary strength of our review is that it, for the first time, provides a systematic synthesis of the existing available literature and offers unique insights into the totality of the evidence concerning the associations between psychological stress and the gut microbiota composition. Despite this, our review has several limitations. These being that there is currently a lack of evidence to enable stratification by age, gender, or ethnicity in a meaningful way. As previously mentioned, these are areas of future research which should warrant attention. Similarly, other factors such as diet and medication use were not considered. Aspects such as these are important since it is well known that dietary factors can strongly modulate the gut microbiota in complex and intricate ways which offer their own challenges in terms of measurement and interpretation which were beyond the scope of this review [46]. Furthermore, various pharmacological agents such as proton pump inhibitors, metformin, and laxatives have also been shown to influence the composition and function of the microbiome and were not considered in this study [47]. Moreover, the available data were largely derived from observational studies as few randomized controlled trials have been conducted in this area. Of the studies which have been conducted only a minority investigate the recovery time of the microbiota. This is important since some bacterial elements, such as the ratio of Fermicutes to Bactereoidetes, may have a relatively good degree of plasticity yet their associated metabolism may not, leading to potentially slower recuperations times from stress-related insults in animals as well as adults [8, 48]. Finally, from the available evidence it is also impossible to determine the direction of the relationship between stress and the microbiota or to infer causality. These are all areas which should be considered for future research and addressing these aspects would further strengthen our understanding of the relationship between psychological stress and the composition of the gut microbiota.

In conclusion, our review of the literature shows for the first time that there is an overall negative association between psychological stress and measures of alpha- and beta diversity of gut microbiota. Psychological stress was associated with a decreased abundance of Proteobacteria and Verrucomicrobia, and increased Euryarchaeota at the phylum level and was also associated with a change in the abundance of microbiota at the genus-level. There is currently a lack of evidence regarding the impact of gender, age, and ethnicity upon these factors and this should be addressed in future research. Larger sample longitudinal studies are needed to provide the causal relationship between psychological stress and gut microbiota.

An ethics statement is not applicable because this study is based exclusively on published literature.

The authors declare that they have no competing interests.

This work was supported by China Medical Board (Grant No. 16-262), National Institutes of Health (Grant No. U54 HD070725), United Nations Children’s Fund (Grant No. Unicef 2018-Nutrition- 2.1.2.3), the Chinese National Key Research and Development Program (Grant No. 2017YFC0907200 and 2017YFC0907201), the National Natural Science Foundation of China (8210120946), Natural Science Basic Research Program of Shaanxi (2020JQ-094), and China Postdoctoral Science Foundation (2019M653669).

L.M., M.M., and Y.W. designed the research; Y.Y., Y.L., and L.M. conducted the literature search, data screening, and extraction; L.M., R.J.W., Y.Y., and S.M. drafted and revised the manuscript; B.X. and X.S. help revised the manuscript; L.M., M.M., and Y.W. provided administrative support for the project and had primary responsibility for the final manuscript; all authors read and approved the final manuscript.

Lu Ma, Yating Yan and Richard James Webb contributed equally to this work.