Expressive writing as a practice against work stress: A literature review

Abstract

Expressive writing is a well-structured written emotional disclosure intervention for processing stressful experiences. Despite its origins in clinical psychology, studies with occupational samples show positive effects on personal resources and well-being. According to the PRISMA guidelines (Moher et al., 2009), a systematic literature review was conducted in Scopus, Web of Science, Business Source Ultimate, and PsycINFO. Thirteen studies were conclusively reviewed. Nine showed significant positive effects on mental health variables (e.g., depressive symptoms), personal resources (e.g., self-efficacy), or organizational variables (e.g., job satisfaction). Based on this, we examined the implementation protocols used, sought to identify critical factors for the effectiveness in an occupational context, and derived a recommendation for evidence-based implementation conditions. Participants benefit from writing four times over a 20-minute period in which they are guided by written instructions to engage with a stressful work-specific experience cognitively and emotionally. However, the effectiveness varies for different groups, especially regarding baseline stress. Expressive writing had a more beneficial effect on individuals with lower resources and higher stress levels. Practical use cases in occupational health and human resource management are discussed. The results are limited in the restricted choice of databases and search terms and the ambiguous definition of the evaluation criteria.

Keywords:

• Expressive writing
• occupational stress
• stress intervention
• stress management
• written emotional disclosure

Introduction

Work stress affects an employee’s health and the performance of the organization they are working for. Research investigating the impact of working conditions on health repeatedly points to the link between stress and negative outcomes, such as burnout or depression (Alarcon, 2011; Aronsson et al., 2017; Theorell et al., 2015). In addition, occupational stress seems to be a significant source of costs. The World Health Organization (WHO) estimates the annual cost of occupational stress at $1 trillion (WHO, 2013). Stress management should be integrated as part of organizational health management to address these negative effects. Stress management interventions that focus on the individual level and address cognitive-behavioral processes seem to affect work stress positively (Holman et al., 2018). These practices in occupational health management could alleviate stress and potentially prevent harmful effects on employee health and organizational productivity. To integrate organizational processes and health management sustainably, interventions that can be implemented cost- and time-efficiently are particularly suitable in the organizational context.

The paradigm of written emotional disclosure might offer an approach for this purpose. Written emotional disclosure can be defined as a well-structured writing exercise that is used to promote health and positively influence psychological outcomes (Caputo et al., 2022). The frequency, duration, and subject matter of writing are specified in the implementation protocol and differ depending on the purpose of the exercise. Therefore, written emotional disclosure basically involves different interventions that vary in their design (e.g., writing a thank-you letter once a week, writing down three things that went well each day, exploring different visions of a desirable future) (Seligmann et al., 2005).

This paper will focus on an intensively researched specific form of written emotional disclosure called expressive writing (Gao, 2022; Pennebaker & Beall, 1986; Pennebaker & Chung, 2011). Expressive writing is an intervention designed to process distressing experiences by writing about one’s deepest thoughts and feelings regarding a stressful life event in three to five sessions of 20 minutes each. These writing sessions are accompanied by instructions in which the assignment is formulated. Usually, the instructions contain a description of the content of the essays, a request to reflect deeply on one’s thoughts and feelings, and a call not to pay attention to spelling and orthography (Gao, 2022; Pennebaker & Chung, 2011).

Although expressive writing originates in clinical psychology (Pennebaker & Beall, 1986), it has also been shown to affect non-clinical populations positively. Expressive writing has been associated with faster reemployment after unemployment (Spera et al., 1994), improved coping with a homosexual orientation (Swanbon et al., 2008), and a more realistic body image perception for women with distorted body image perceptions (Lafont & Oberle, 2014). In addition, several studies also indicate that expressive writing in a professional context can have a positive influence on health variables (e.g., depressive symptoms) (Procaccia et al., 2021), personal resources (e.g., self-efficacy, resilience) (Kirk et al., 2011; Saldanha & Barclay, 2021), as well as organizational variables (e.g., work engagement, continuance commitment) (Cosentino et al., 2021; Lanaj et al., 2019).

Despite these recurrent positive effects of expressive writing, meta-analytic results examining clinical and non-clinical populations are somewhat inconsistent. Although slight to moderate impacts were found in most of the conducted meta-analyses (Frattaroli, 2006; Frisina et al., 2004; Pavlacic et al., 2019; Qian et al., 2020; Smyth, 1998), three meta-analytical studies were not able to detect an effect (Meads & Nouwen, 2005; Mogk et al., 2006; Reinhold, Brückner & Holling, 2018). Reinhold et al. (2018) assumed that moderating variables related to both the individual (e.g., individuals with high-stress levels seem to benefit more than individuals with lower levels) as well as the intervention protocol (e.g., longer writing sessions appear to be more effective than shorter ones) impact the effect of the writing intervention. Therefore, variables that can potentially moderate the relationship between expressive writing and its outcomes should also be investigated in the occupational field of application. A recent meta-analysis by Caputo et al. (2022) examining the impact of expressive writing on work stress found incongruent effects. For work stress, they did not find a significant positive effect of the intervention. For variables of personal well-being, however, they found a significant medium-sized effect. The authors integrate their results into the current state of research and refer to explanatory approaches by Harris (2006). Harris (2006) assumes that expressive writing has limited effects when it is used to counter a specific stressor (e.g., unfair treatment at work). However, the authors argue that the low cost and time effectiveness of expressive writing should be reasons to implement the intervention despite its small effects.

Three theoretical approaches explain the basic mechanisms of expressive writing: inhibition theory, habituation, and narrative construction (Gao, 2022; Pennebaker & Chung, 2011). First, inhibition theory guided early work on expressive writing (Pennebaker, 1989; Watson & Pennebaker, 1989). Its basic assumption is that inhibited thoughts and feelings lead to physical strain since continuous psychological effort is required to suppress emotions and thoughts. The continuous suppression provides mild but constant interference to the autonomic and central nervous systems. Releasing this inhibition through expressive writing relieves the strain so that the observed health benefits occur. Empirical evidence for this theory can be found in the work of Swanbon et al. (2008). They experimentally examined the influence of expressive writing on avoiding one’s homosexual orientation and associated somatic complaints. Results showed that participants who wrote about their deepest thoughts and feelings regarding their sexual identity exhibited less avoidance, rumination, and somatic complaints than participants who wrote about a neutral topic.

In the second theoretical approach, the effect of expressive writing is explained by habituation. In line with this view, repeated exposure to stressful thoughts and feelings should trigger a habituation process (Rankin et al., 2009; Sloan et al., 2005; Sloan & Marx, 2004). Researchers supporting this claim explain the effectiveness of the intervention based on Emotional Processing Theory (e.g., Foa et al., 2007; Foa & Kozak, 1986). This theory proposes that by the emotional exposure within the writing sessions, memories lose their emotional intensity, anticipated negative consequences do not occur, and new, less stressful cognitive patterns can form. Sloan et al. (2005) showed that repeated writing about the same traumatic experience had more significant health effects than expressive writing about different traumas. From this finding, the authors deduced that more frequent writing equates to greater exposure, promoting habituation.

Research that focused primarily on the content of the resulting essays provided a third explanation: narrative construction. Via expressive writing, thoughts are sorted, and a coherent narrative is formed. With this cognitive achievement, thoughts and feelings can be efficiently integrated into the memory and do not require further regulation resulting in improved well-being (Pennebaker, 1993; Pennebaker & Francis, 1996). A recent study of divorced adults by Bourassa et al. (2017) provides empirical evidence for this third explanation. It shows that guiding participants to formulate a coherent narrative in their essays can positively impact cardiac health. Participants in the narrative condition had significantly lower average heart rates and higher heart rate variability than the group that performed the intervention without instruction to form a coherent narrative and the control group. Whether the effect of expressive writing can be explained based on one of these three theoretical approaches remains questionable. Pennebaker and Chung (2011) assume that all three processes work simultaneously, interact, and influence each other.

To our knowledge, these theoretical assumptions are yet to be tested in an organizational context. To theoretically justify the efficacy of expressive writing for occupational stress, one could also apply the transactional stress theory (Lazarus & Folkman, 1987) or the Job Demand-Resource Model (JD-R) (Demerouti et al., 2001; Bakker & Demerouti, 2014; Bakker & Demerouti, 2017).

In the transactional stress theory, stress is understood as the result of an appraisal process that concludes that external demands exceed one’s resources and thus represent a threat (e.g., the threat to self-esteem, social belonging, or competence) to the individual, which produces stress. In the primary appraisal process, a stressor is initially evaluated as a challenge, a threat, or irrelevant. The secondary appraisal process takes place simultaneously and does not refer to the stressor itself but to the resources available to the individual to address this stressor. Stress arises if the available resources are assessed as too low (Lazarus & Folkman, 1987). The JD-R Model is based on a similar premise. Here, too, it is assumed that stress arises when the demands of a situation exceed the available organizational (e.g., latitude, social support, feedback) and personal resources (e.g., social skills, expertise).

If the effect of expressive writing is explained based on these two theoretical approaches, the intervention could serve two functions: In terms of the JD-R Model, expressive writing could be a practical behavioral resource for employees (Bakker & Demerouti, 2014; Bakker & Demerouti, 2017). This expansion could positively contribute to the buffering effect of resources against demands and therefore protect against work stress.

In line with the transactional stress theory, self-reflection within the expressive writing sessions could be a conscious reappraisal process. The session provides enough space to consider resources that may have been inaccessible at first. For example, employees could reflect on their demands, question the significance of events, and recall situations in which similar challenges were overcome.

Based on the broad research on expressive writing, this paper will examine how expressive writing can be used to address occupational stress. For this purpose, the existing research is summarized in a literature review. A particular focus is placed on previous studies’ implementation conditions and intervention protocols to develop well-founded implementation guidelines in an organizational setting. These guidelines should ensure that participants receive the most positive impact from the intervention. Furthermore, this review article serves the scientific community by summarizing and integrating parts of the written emotional disclosure paradigm in an applied setting and providing a practical framework for using expressive writing within the workplace.

The relevant literature is first identified and then examined for internal validity. In the following step, the results of the relevant studies are presented, and the implementation protocols are reviewed. In the discussion, guidelines for designing expressive writing interventions will be deduced, and limitations and different research approaches will be discussed.

Research Methodology

A process-oriented approach is taken for conducting the review. Thus, antecedents, consequences, and limitations of expressive writing in the professional context are elaborated. The PRISMA method made the literature review as transparent as possible (cf. Moher et al., 2009, see Figure 1). Therefore, inclusion criteria are defined first, followed by the presentation of the identification, screening, eligibility, and inclusion phases of the selection process. The literature search was conducted in November 2021 and extended between 23 February 2023, and 10 March 2023. The results of the searches are presented below. The lead author conducted the literature searches. The screening process was also reviewed by two independent researchers

Figure 1. Representation of the literature selection process based on the PRISMA method (cf. Moher et al., 2009).

Inclusion criteria

Two inclusion criteria can be derived directly from the research question: Since the purpose is to evaluate how expressive writing can be used as an effective intervention to reduce occupational stress, only studies that both comprise an occupational sample and use expressive writing will be included. In addition, a relationship with psychological well-being or reduction of occupational stress must be evident. Standard operationalization is not required. Studies investigating well-being constructs as dependent variables (e.g., resilience, burnout, coping strategies, etc.) will be considered. Only experimental or quasi-experimental designs that are published in a peer-reviewed journal are included to ensure the greatest possible conclusiveness and quality of the results. Due to the language skills of the researchers, included papers must be published in German or English.

In summary, the following inclusion criteria were defined:

• The effectiveness of expressive writing is examined in an occupational sample,

• the dependent variable can be operationalized or is related to occupational stress or psychological well-being,

• an experimental or quasi-experimental design was used,

• the study has been published in a peer-reviewed journal and

• the study is published in German or English.

Identification of literature

The literature was identified from four databases: Scopus, Web of Science, Business Source Ultimate, and PsycINFO. Since the research question is interdisciplinary, it was decided to examine two discipline-independent databases (Scopus and Web of Science), one that focuses more on organizational management literature (Business Source Ultimate) and one that compiles psychological literature (PsycINFO).

For each combination of search terms, one search round was performed. One round includes entering the search term combination into one of the databases and screening the results. A total of 58 rounds were conducted. Two terms were used in each round. One was standardly defined as “expressive writing” or “written emotional disclosure”. The second one was varied as follows: “profession,” “occupation,” “job,” “employ*,” “burnout,” “occupational stress,” and “workplace adversity.” The search terms were linked with the Boolean operator “and” and were searched for in the abstract and title. This procedure resulted in 453 hits. These were distributed among the databases as follows: Scopus: n = 168, Web of Science: n = 179, Business Source Ultimate: n = 15, PsycINFO: n = 91. In the next step, 166 duplicates were removed. These were publications that appeared with more than one search term or in more than one database.

Screening of literature

After removing duplicates, 287 papers were included in the screening process. The hits displayed were checked for inclusion criteria based on the title and abstract during the screening. Two hundred six papers were excluded for the following reasons: the paper was out of scope (n = 180), an occupational sample was not studied (n = 35), expressive writing was not used (n = 6), the key dependent variables were not related to occupational stress (n = 2), a very specific occupational sample was studied (e.g., soldier or athletes) (n = 3), no experimental design was conducted (n = 8), the paper was not peer-reviewed (n = 21), the publication was not written in English or German (n = 5). Thus, the selection was mainly based on the predefined inclusion criteria. The only additional criterion was that very specific occupational groups, such as soldiers or athletes, were excluded. This posthoc exclusion was justified because work stress in these occupational groups includes aspects that are the exception in the broader population (e.g., traumatic events, high physical demands).

The two independent researchers each received half of the data set of the 287 identified articles (researcher 1 received 144 articles and researcher 2 received 143). Each of them had the task of screening the identified articles with respect to the defined inclusion criteria. The selection of the lead author was not transparent to the independent researchers. Thus, each article was screened by both the lead author and an independent researcher. The two screening processes were then compared for concordance. A total of eight articles had discrepancies in screening. For these articles, consensus was reached in an oral debriefing.

Eligibility

The screening process identified 27 articles that were subjected to full-text analysis. This decision was made because it was impossible to assess whether the inclusion criteria were met on the title or abstract. After this full-text analysis, 14 records were excluded for the following reasons: No occupational sample was studied (n = 1), expressive writing or written emotional disclosure was not used (n = 4), a very specific occupational sample (e.g., soldiers or athletes) (n = 1) was investigated, no experimental design was conducted (n = 8). Thirteen articles met the inclusion criteria and were therefore included in the analysis.

Studies included in the review

Thirteen articles were evaluated in this review (Ashley, O’Connor, & Jones 2013; Barclay & Saldhanha, 2016; Barclay & Skarlicki, 2009; Cosentino et al., 2019; 2021; Kirk et al., 2011; Michailidis & Cropley, 2019; Procaccia et al., 2021; Round et al., 2022; Saldanha & Barclay, 2021; Tarquini Trani, & Solano, 2016; Tonarelli et al., 2017; 2018). Ten of the thirteen studies were randomized control trials (Ashley et al., 2013; Barclay & Saldanha, 2016; Barclay & Skarlicki, 2009; Cosentino et al., 2021; Kirk et al., 2011; Michailidis & Cropley, 2019; Procaccia et al., 2021; Round et al., 2022; Saldanha & Barclay, 2021; Tonarelli et al., 2017) and three were quasi-experimental designs (Cosentino et al., 2019; Tarquini et al., 2016; Tonarelli et al., 2018). The selected studies were all peer-reviewed and examined in a broad professional context. In addition, they all investigated the influence of an expressive writing protocol on variables that can be attributed to psychological well-being.

Evaluation process

The selected studies are assessed on two levels: First, the study design is investigated regarding methodological quality, followed by a review of the substantive results.

Internal validity is used as an evaluation criterion for methodological quality. Internal validity represents the trustworthiness of the effect produced by the manipulation in an experimental design (Campbell, 1957). Campbell (1957) defines seven validity threats that can negatively affect this trustworthiness (history, maturation, testing, instrument decay, regression, selection, and mortality). Evaluating these threats can serve as a basis for assessing internal validity (Campbell, 1957).

This approach is applied in the following analysis. The thirteen selected studies are assessed in terms of their internal validity threats so that the trustworthiness of the study design can be assessed. Flannelly et al. (2018) and Lander (1998) provide a link between different experimental designs and these threats. For experimental and quasi-experimental studies, they consider maturation, selection, and mortality to be particularly important. Based on this, maturation, selection and mortality validity are assessed for each study. In evaluating the quasi-experimental studies, the validity threat of the history is also considered. In non-randomly assembled groups, the history preceding the study may theoretically systematically bias the manipulation and is therefore assessed (Flannelly et al., 2018).

An analysis of the substantive results follows. For this purpose, the most important aspects of the study are presented, including the characteristics of the sample, the main dependent and independent variables, the protocol used, and the main results.

Results

The following section presents the results of the internal validity assessment of all selected studies. Subsequently, the results of the studies are summarized based on the methodology of the expressive writing protocol and the differentiation therein, as well as the positive and unexpected effects.

Assessment of internal validity

Table 1 summarizes the results of the internal validity assessment. First, the assessment for the ten randomized control trials is presented, followed by an evaluation of the quasi-experimental studies.

Table 1. Tabular comparison of the internal validity threats of the studies analyzed in the review.

Article listing by researcher	Comparison of internal validity threats
	Maturation	Selection	Mortality	History
Ashley et al. (2013)	low threat	very low threat	threat	–
Barclay and Saldanha (2016)	low threat	low threat	low threat	–
Barclay and Skarlicki (2009)	low threat	low threat	very low threat	–
Cosentino et al. (2019)	low threat	threat	very low threat	threat
Cosentino et al. (2021)	low threat	very low threat	very low threat	–
Kirk et al. (2011)	increased threat	low threat	low threat	–
Michailidis Cropley (2019)	increased threat	very low threat	increased threat	–
Procaccia et al. (2021)	low threat	very low threat	very low threat	–
Round et al. (2022)	low threat	very low threat	threat	–
Saldanha and Barclay (2021)	low threat	low threat	increased threat	–
Tarquini et al. (2016)	threat	threat	very low threat	intense threat
Tonarelli et al. (2017)	low threat	very low threat	very low threat	–
Tonarelli et al. (2018)	low threat	threat	very low threat	threat

Randomized control trials

Flannelly et al. (2018) proposed maturation, selection, and mortality as threats to internal validity in experimental studies. Maturation describes the influence of time in an experimental study. When research takes a certain amount of time, participants are changed by influences that cannot be controlled experimentally. The effects of these changes cannot always be estimated (Campbell, 1957). This effect can be counteracted, if a study conducts a pre- and post-comparison. The closer the post-measurement is to the end of the intervention, the easier it is to exclude maturation influences. In addition, the experimental and control groups should receive the treatment simultaneously. Otherwise, the validity threat due to maturation increases (Flannelly et al., 2018).

All ten studies conducted a pretest and had the experimental and control groups perform the intervention simultaneously. Therefore, maturation can be predominantly assessed as a very low threat. In two studies, however, a limitation is made: the post-measurement of Kirk et al. (2011) and Michailidis and Cropley (2019) did not directly follow the end of the intervention. In the study by Kirk et al. (2011), it was two weeks after the end of the writing sessions, while Michailidis and Cropley (2019) waited for four weeks. The authors’ approach could be justified because this measurement allows for a more long-term assessment of the effects of expressive writing. However, considering this argument, a pre-, post-, and follow-up measurement could serve this purpose and reduce the maturation threat. Although Michailidis and Cropley (2019) have already chosen this procedure, it is recommended to perform the post-measurement directly at the end of the intervention. In this way, the validity threat of maturation could be further prevented.

The second validity threat Flannelly et al. (2018) cited is selection bias. This threat occurs when the selection of participants is influenced by unknown variables (Campbell, 1957). Randomization counteracts this bias. Furthermore, inclusion and exclusion criteria can be defined. If it is theoretically justifiable that certain variables systematically influence the study’s outcome, then these variables can be sorted out in advance. This elimination helps to prevent confounding effects (Flannelly et al., 2018).

In all ten studies, both randomization and inclusion criteria are in place. Most inclusion criteria were directly related to the subject of the study. Therefore, subjects had to work in a certain occupational field (e.g., medical field) (Cosentino et al., 2021; Procaccia et al., 2021; Tonarelli et al., 2017) or had to have certain experiences (e.g., experiences of injustice) (Barclay & Saldanha, 2016; Barclay & Skarlicki, 2009; Saldanha & Barclay; 2021). Ashley et al. (2013), Cosentino et al. (2021), Procaccia et al. (2021), Round et al. (2022), and Tonarelli et al. (2017) further differentiated the employment relationship. They determined a minimum number of hours worked per week and a minimum length of employment, as it can be argued that these factors influence the experience of occupational stress. Ashley et al. (2013), Michailidis and Cropley (2019), and Round et al. (2022) defined exclusion criteria as well. Ashley et al. (2013) and Round et al. (2022) excluded people with psychiatric clinical diagnoses from the outset because they suspected that these diagnoses might be a potential confounding variable for work stress. Michailidis and Cropley (2019) differentiated even more precisely. They assumed that increased scores on the posttraumatic embitterment disorder scale (Linden et al., 2008) systematically influence the effect of expressive writing. Therefore participants with high scores on this scale were excluded (n = 73). This approach is particularly beneficial regarding internal validity, as it prevents systematic distortions. The studies by Ashley et al. (2013), Cosentino et al. (2021), Michailidis and Cropley (2019), Procaccia et al. (2021), Round et al. (2022) and Tonarelli et al. (2017) are therefore rated lowest in terms of threat from the selection. Since the other studies did not include any additional reflection of possible interfering influences but selected a sample specific to the subject of the study, the assessment for validity threats due to selection bias is also rated as low.

Mortality describes the exit of participants from an experiment (Campbell, 1957). It threatens internal validity when dropout varies between experimental and control groups (Flannelly et al., 2018). Four of the ten studies had no dropouts at all and thus no threat in terms of mortality (Barclay & Skarlicki, 2009; Cosentino et al., 2021; Procaccia et al., 2021; Tonarelli et al., 2017). Barclay and Saldanha (2016) and Kirk et al. (2011) had a very small dropout, which was almost equally distributed in both groups. Therefore, no significant validity threat is assumed either. The studies by Ashley et al. (2013) (n = 49), Michailidis and Cropley (2019) (n = 18), Round et al. (2022) (n = 38), and Saldanha and Barclay (2021) (n = 13) had larger dropout rates. Therefore, the threat of mortality is rated higher for these studies than the others. However, further differentiation is made between the studies of Michailidis and Cropley (2019), Saldanha and Barclay (2021), Ashley et al. (2013), and Round et al. (2022). Comparatively, fewer participants dropped out of the studies by Michailidis and Cropley (2019) as well as by Saldanha and Barclay (2021). The participants who dropped out were evenly distributed between the experimental and control conditions. Therefore, mortality in these designs is rated as an increased threat. The research design by Ashley et al. (2013) and Round et al. (2022) is rated as threatened because the dropout rates are considerably higher and are not evenly distributed across the conditions.

Quasi-experimental studies

Cosentino et al. (2019) and Tonarelli et al. (2018) use a classic quasi-experimental comparison of a control and an experimental group without randomization. The study by Tarquini et al. (2016) is a nonequivalent control group design. Here, two groups are compared, which were not randomized. One group received treatment, and the other did not (Cook & Campbell, 1979). For a differentiated assessment of internal validity in the quasi-experimental studies, the three threats of maturation, selection bias, and mortality are discussed, and history is also considered.

In the case of maturation, the pre-and post-measurements can address the validity threat because preexisting differences are recorded in the pre-measurement. All three studies conducted a baseline survey and executed the post-tests directly after the last session of the intervention. Nevertheless, the Tarquini et al. (2016) study is susceptible to a validity concern specifically related to maturation. This is because the changes observed in the comparison group were neither manipulated nor controlled, making it difficult to accurately estimate their impact.

Also, the selection biases for all three studies can be assumed. Since participants are not randomly assigned, it is more likely that differences between groups are already present at baseline (Flannelly et el., 2018). Instead of a control group, it is, therefore, more appropriate to refer to a comparison group. Cosentino et al. (2019) attempted to reduce selection bias with block randomization. Tonarelli et al. (2018) also attempted to address this threat to validity with a balanced sampling for organizational settings and years of work experience. However, both studies showed significant differences between study groups. The study by Cosentino et al. (2019) showed significant differences in normative commitment in the baseline measurement. Tonarelli et al. (2018) study showed differences in avoidance and satisfaction of colleague relationships. Tarquini et al. (2016) indicated that their comparison and experimental groups did not differ significantly on sociodemographic characteristics or dependent variables. However, this does not mean that there is no selection bias. In the Tarquini et al. (2016) study, although the groups did not vary based on the variables presented, there was still a systematic bias. In this study, the assignment was based on place of residence. Accordingly, it is reasonable to assume that living conditions in one city compared to another systematically influence participants.

The experimental design does not directly affect mortality (Flannelly et al., 2018), and no dropouts were reported (Cosentino et al., 2019; Tarquini et al., 2016; Tonarelli et al., 2018). Therefore, mortality is very low in the three quasi-experimental studies.

In quasi-experimental studies, Flannelly et al. (2018) also discussed the validity of threat history. This refers to systematic biases that occur because of historical events. Cosentino et al. (2019) and Tonarelli et al. (2018) examine workgroups, and Tarquini et al. (2016) grouped participants based on their place of residence. In both group assignments, history bias can be assumed. Both workgroups and communities have histories of their own and perceive historical events in a particular context. Therefore, this threat is classified as given. However, this threat is estimated to be highest in the study by Tarquini et al. (2016), where no balancing of the groups had occurred.

Key Findings

Table 2 shows the research results based on the research subject, sample characteristics, the key-dependent and independent variables, and the key findings of all thirteen studies included in the review. In the following section, these studies will be summarized and compared.

Table 2. Tabular comparison of the studies analyzed in the review.

Article listing by researcher	Research subject	Design	Sample characteristics	Key dependent variable	Independent variable	Key findings
Ashley et al. (2013)	Evaluation of written emotional disclosure interventions on the self-reported health and job satisfaction of school teachers	Randomized control trial	Participants were recruited through advertisements in schools in the UK,5.33 years of work experience, 43.62 average age, 85.71 % females (N = 77)	Psychological health, Physical health, Job satisfaction	Vriations of expressive writing protocol	No significant difference could be found between experimental groups and the control group with respect to the dependent variables Physical health, however, improved in all writing conditions (experimental and control); this improvement was still evident six months after the intervention
Barclay and Saldanha (2016)	Evaluation of the relationship between expressive writing and forgiveness after an injustice at work	Randomized control trial	Participants were recruited on a North American university campus,5.33 years of work experience, 20.85 average age, 67.7 % females (N = 96)	Overall Injustice Perceived Resolution Forgiveness	Experiences with injustice in the workplace and variations of the expressive writing protocol	Expressive writing was associated with higher levels in perceived resolution meditated by positive emotions and perceived injustice Perceived resolution also mediated the relationship between expressive writing and forgiveness
Barclay and Skarlicki (2009)	Evaluation of the effectiveness of expressive writing after an injustice at work	Randomized control trial	Participants were recruited on a North American university campus,5 years average of work experience, 23 average age, 75 % females (n = 100)	Physical Symptoms Psychological Well-being Anger Retaliation Intentions Perceived Resolution	Experiences with organizational injustice	Expressive writing was associated with higher psychological well-being, fewer intentions to retaliate, and higher levels of personal resolution
Cosentino et al. (2019)	Evaluation of the expressive writing protocol in palliative care	Pretest–Posttest Control Group Design (no randomization)	Palliative care staff from Northern Italy, 34.8 %; were between 26-35 years old, 68.2 % females (N = 66)	Compassion Satisfaction, Group Cohesion, Organizational Commitment	Expressive writing protocol	The analysis did not show significant differences between the experimental group and control group
Cosentino et al. (2021)	Evaluation of the expressive writing protocol in palliative care	Randomized control trial	Palliative care staff from Northern Italy, 44.74 average age, 78 % females (N = 50)	Compassion Satisfaction, Organizational Commitment, Resilience, Compassion Fatigue Perceived Stress	Expressive writing protocol	Expressive writing was associated with higher continuance Commitment, lower sleep disturbances, less anger, less intrusiveness, less hyperarousal
Kirk et al. (2011)	Evaluation of expressive writing protocol in terms of self-efficacy, emotional intelligence, positive affect and workplace incivilities	Randomized control trial	Various personnel from New South Wales and Queensland Australia, 3.1 years average work experience, 35.1 average age, 67.35 % females (N = 49)	Emotional Self-efficacy Emotional Intelligence Positive Affect Negative Affect Incivility Victimization Incivility Perpetration	Experiences incivility in the workplace, expressive writing protocol	Expressive writing was associated with increase in self-efficacy for participants with lower pretest scores, increase in emotional intelligence, positive affect, and decrease in workplace incivility perpetration
Michailidis and Cropley (2019)	Evaluation of the influence of expressive writing on workplace embitterment	Randomized control trial	Various personnel from the UK, 45 % have worked in their current role 1-3 years, average age 34.22, 50 % females (N = 44)	Embitterment, Work-related Rumination, Sleep Quality	Expressive writing protocol	No significant differences were found between the control and experimental group. In both groups, embitterment and affective rumination decreased, detachment increased, and sleep quality improved
Procaccia et al. (2021)	Evaluation of the influence of expressive writing on psychological adjustment of healthcare workers during the Covid Pandemic	Randomized control trial	Palliative care staff from Italy, 46.42 average age, 74.54% % females (N = 55)	Perceived social support, Resilience Depression PTSD	Expressive writing Protocol	Expressive writing was associated with a decrease in PTSD, depression and global psychopathology In addition, the effect of expressive writing was found to depend on demographic variables
Round et al. (2022)	Examining the influence of an expressive writing protocol that encourages positive reflection on events on teachers and non-teachers	Randomized control trial	Teachers and non-teachers from the UK 38.1 average age, 81.28% % females (N = 66)	State anxiety Burnout Job satisfaction Perceived stress Trait anxiety Physical health	An expressive writing protocol that encourages intensive reflection on a particularly positive experience	State anxiety and job satisfaction were positively influenced by the experimental condition of positive expressive writing There was no difference between teachers and non-teachers
Saldanha and Barclay (2021)	Validate a new expressive writing intervention that promotes meaning-finding to help deal with workplace unfairness	Randomized control trial	Participants were recruited on a North American university campus, 3.34 years of work experience, 20.32 average age, 43.4 % females (N = 131)	Willingness to reconcile Positive relationships with others Resilience Situation Resolution	Experiences with unfairness in the workplace and variations of the expressive writing protocoll	The meaning-finding protocol is associated with higher resilience than traditional expressive writing Resilience has a mediating effect on the relationship between expressive writing and willingness to reconcile, positive relationships with others, and life satisfaction.
Tarquini et al. (2016)	Evaluate the impact of expressive writing on occupational burnout.	Non equivalent Control Group Design	Italian administrators from Siena and Chianciano Terme, 49.94 average age, 51.43 % females (N = 35)	Exhaustion, Cynicism, Professional Ineffectiveness Alexithymia Difficulty Identifying Feelings Difficulty Describing Feelings Self-acceptanc Positive Relations With Others Externally-oriented Thinking Environmental Mastery Personal Growth Dimension	Expressive writing protocol	Expressive writing was associated with a decrease of all three burnout dimensions, alexithymia, and an increase in environmental mastery, self-acceptance and positive relations with others
Tonarelli et al. (2017)	Evaluate the impact of an expressive writing protocol on coping strategies, emotions and relationships, and on work satisfaction of health care professionals	Randomized control trial	Healthcare staff from Bologna, 45.47 average age, 97 % females (N = 66)	Coping Strategies, Working Satisfaction Relational Communication Satisfaction	Expressive writing protocol	In the follow-up measurement, a rebound effect was found in the avoidant coping style
Tonarelli et al. (2018)	Review of findings previously obtained in 2017 on the relationship between expressive writing and coping with occupational stress in the field of palliative care	Pretest–Posttest Control Group Design (no randomization)	Palliative care staff from Italy 46.15 average age, 88,5% female (N = 27)	Avoidance Transcendent Positive aptitude Social support Problem orientation Emotional exhaustion Depersonalization Personal fulfillment Explicit recognition Work-family balance Work family Colleagues Social interaction Professional opportunities Praises and recognitions Responsibilities control	Expressive writing protocol	The experimental group showed lower avoidant coping strategies and improved work-colleague relationships.

This paper examined thirteen studies, all of which investigated the influence of expressive writing on various variables associated with psychological well-being in an occupational sample. The number of participants ranged from N = 27 to 131. Women formed the majority in twelve of the thirteen samples. The dependent variables examined were predominantly related to facets of psychological well-being (e.g., depression, anger, etc.). Additionally, six studies reviewed organizational outcomes (e.g., continuance commitment, job satisfaction) (Ashley et al., 2013; Cosentino et al., 2021; Kirk et al., 2011; Round et al., 2022; Tonarelli et al., 2017; 2018).

Positive effects of expressive writing

The positive effects of expressive writing are summarized in Figure 2. Nine of the thirteen studies found a significant beneficial effect of expressive writing. Expressive writing had a positive influence on mental health variables such as burnout (Tarquini et al., 2016), depression, PTSD (Procaccia et al., 2021), and state anxiety (Round et al., 2022) as well as sleep disturbances (Cosentino et al., 2021). Expressive writing was most effective for participants with greater discomfort. For example, in the Procaccia et al. (2021) study, there was an interaction effect between the expression of psychopathological symptoms (e.g., PTSD, depression) in the pretest and the effectiveness of the intervention. This means that participants who reported being more affected were also more likely to benefit from the intervention.

Figure 2. Graphical representation of the positive effects of expressive writing depending on the study design and intervention protocol.

A similar trend emerged regarding the positive effects of expressive writing on personal resources. Improvements could be observed in self-efficacy (Kirk et al., 2011), resilience (Saldanha & Barclay, 2021), and self-acceptance (Tarquini et al., 2016). Kirk et al. (2011) found that individuals with low or moderate self-efficacy benefited significantly from expressive writing. No significant effect was found for participants with high self-efficacy. Again, it appears that the impact of expressive writing is strongest for individuals with greater limitations (due to fewer personal resources).

It appears that improvement in personal resources through expressive writing is also related to improved social relationships. In the study by Saldanha and Barclay (2021), resilience was increased and functioned as a mediating variable for the relationship between expressive writing and reconciliation, positive relationships with others, and life satisfaction, Cosentino et al. (2021) found a positive influence on help-seeking behavior, and Tonarelli et al. (2018) showed a positive impact of the intervention on the assessment of relationships with colleagues.

Not only did variables of personal well-being improve, but there was also a tendency for positive developments in organizational variables. Cosentino et al. (2021) showed significantly higher continuation commitment in the experimental group, Tonarelli et al. (2017) and Round et al. (2022) observed improved job satisfaction, and research by Barcley and colleagues (Barclay & Saldanha, 2016; Barclay & Skarlicki, 2009; Saldanha & Barclay, 2021) provides evidence for the use of intervention in coping with organizational injustice.

The positive effects of expressive writing are also found in studies with a high internal validity estimate. Seven of the nine studies finding positive effects are randomized control trials with a small estimate of internal validity threats.

Unexpected effects of expressive writing

In addition to the positive effects of expressive writing, unexpected effects were also found. Four of the thirteen studies show no or conflicting results. Cosentino et al. (2019) found no significant effect between the control and experimental groups. Michailidis and Cropley (2019) showed that the control and experimental groups reduced embitterment and affective rumination and improved detachment and sleep quality. Similar results were found by Ashley et al. (2013). In their study, expressive writing had no significant effects on the studied variables, but self-rated physical health improved in all writing conditions. This effect was still observed six months after the intervention. Tonarelli et al. (2017) found a rebound effect. In the follow-up measurement seven months after the intervention, avoidant coping styles increased significantly in the experimental group.

Cosentino et al. (2019) justify the results of their work based on the study design. They refer to the number of sessions conducted in the expressive intervention and the number of dependent variables. They hypothesize that significant effects would have occurred if they had conducted more sessions and collected more variables. This hypothesis is supported by the significant results of a similar team of researchers and a similar research plan from 2021 (Cosentino et al., 2021). Here, the protocol was more closely adapted from Pennebaker’s (1994) original one, and two additional dependent variables were collected. In this study, as suspected, expressive writing significantly affected the dependent variables of continuance commitment, sleep quality, anger, and help-seeking.

In the study by Michailidis and Cropley (2019), embitterment, affective rumination, detachment, and sleep quality were improved in both the experimental and control group. This result is particularly noteworthy given that embitterment appears to be a stable construct (Linden et al., 2008; Michailidis & Cropley, 2018). The authors discuss their findings in terms of two aspects: They question the influence of self-distance on embitterment and the time after adversity. Like the experimental group, participants in the control group wrote for 20 minutes on three consecutive days but about their daily routine instead of their thoughts and feelings. This exercise may have led participants to distance themselves from their embitterment, improving their well-being. The authors refer to Park, Ayduk, and Kross’ (2016) research, showing that expressive writing can lead to self-distancing. It is also possible that expressive writing did not achieve the desired effect because the triggering incident was too close to the intervention. The authors did not control this variable and, therefore, cannot provide any information on its influence.

Ashley et al. (2013) discuss stress levels at baseline and their used measures of psychological well-being for the unexpected effects. They note that the sample had good mental health at baseline. Therefore, the authors hypothesize that the measurement was not sensitive enough to possible small changes in an already good mental health well-being and thus could not have resulted in significant effects. They also discuss sample size as a possible cause. In this context, attention should also be drawn to the study’s high dropout (n = 49) of the study, which could limit the internal validity and, thus, the manipulation of the effect. Similar to Michailidis and Cropley (2019), the study by Ashley et al. (2013) also showed unexpected improvements in health variables across all writing conditions. The authors discuss various explanations for this (e.g., historical effects, seasonal variation, regression to the mean) but have yet to reach a clear conclusion. They suggest moderators of health improvement that still need to be fully identified. In the Ashley et al. (2013) study, distancing from daily life while writing also triggered a positive health effect, regardless of the content of the essays.

Tonarelli et al. (2017) hypothesized that expressive writing would increase problem-oriented coping. However, after a non-significant reduction in the post-measurement, the avoidant coping style increased significantly in the follow-up measurement. The authors explain this finding concerning the theory of emotional exposure (Rauch & Foa, 2006). Based on this theory, they assume that the adversity addressed in the intervention could not yet be sufficiently reconstructed so that no habituation effect can occur. Stressors were thus revealed during the writing intervention but not conclusively dealt with. They suspect this produced the observed paradoxical effect (Tonarelli et al., 2017).

Concerning the unexpected results, it should be noted that increased threats to internal validity were identified for three of the four studies with unexpected effects. The study by Cosentino et al. (2019) is quasi-experimental, thus exhibiting a greater selection bias. Since the participants are colleagues, a threat from history cannot be excluded. The study by Michailidis and Cropley (2019) sets its post-measurement at four weeks after the end of the intervention and is, therefore, vulnerable to maturation. In addition, the study has an increased dropout rate (n = 18), just like the research by Ashley et al. (2013) (n = 49).

Differences in expressive writing protocol

Table 3 displays the individual writing protocols of each included study. In each study, two to five sessions of expressive writing were conducted. The median number of sessions was three. In eleven of the thirteen studies, participants wrote for 20 minutes each. The research team led by Cosentino et al. (2019, 2021) chose sessions of 15 minutes.

Table 3. Tabular comparison of intervention protocols used for expressive writing.

Article listing by researcher	Variables of the expressive writing implementation protocol
				Instructions
	Number of sessions	Duration (minutes)	Time between sessions	Experimental group	Control group
Ashley et al. (2013)	3	20	Consecutive	Writing about thoughts and feelings (EG1) / Writing about thoughts and feelings about one single event (EG2) / Writing about thoughts and feelings about a stressful work-related topic (EG 3)	Writing about time management
Barclay and Saldanha (2016)	4	20	Consecutive	Writing about thoughts and feelings about an experience of injustice in the workplace (EG1) / Writing about thoughts and feelings about an experience of injustice in the workplace and forgiveness (EG 2 + EG 3)	Writing about time management, bedroom, daily routine, experimental room
Barclay and Skarlicki (2009)	5	20	Consecutive	Writing about thoughts and feelings about an experience of injustice in the workplace	Writing about time management, bedroom, daily routine, experimental room
Cosentino et al. (2019)	2	15	Time span of 1 – 3 Days	Writing about thoughts and feelings about traumatic, stressful and emotionally significant event in personal life	Write in an objective way about an event that has occurred without dwelling on emotions, thoughts and feelings
Cosentino et al. (2021)	3	15	Time span of 15 Days	Writing about thoughts and feelings about traumatic, stressful and emotionally significant event in personal life	Use of a comparison tool to describe an event as objectively as possible
Kirk et al., (2011)	3	20	Consecutive	Writing about thoughts and feelings about the workday or a significant event at work	Writing about daily routine
Michailidis &Cropley (2019)	3	20	Consecutive	Writing about deepest thoughts and feelings about work	Writing about daily routine
Procaccia et al. (2021)	3	20	Consecutive	Writing about thoughts and feelings about an emotional events	Write in an objective way about an event that has occurred without dwelling on emotions, thoughts and feelings
Round et al. (2022)	3	20	Consecutive	Writing about thoughts and feelings surrounding intensely positive previous experiences	Writing about plans for the rest of the day, shoes they were wearing and their bedroom
Saldanha and Barclay (2021)	4	20	Consecutive	Writing about emotions and thoughts about an experience of injustice in the workplace (EG1) / Writing about thoughts and feelings about an experience of injustice in the workplace and the benefits/positive aspects (EG 2 + EG 3)	Writing about time management, bedroom, daily routine, experimental room
Tarquini et al. (2016)	3	20	Time Span Of 15 Days	Writing about deepest thoughts and feelings about relocation and relating these with their personal situation	–
Tonarelli et al. (2017)	3	20	Consecutive	Writing about thoughts and feelings about traumatic, stressful and emotionally significant event in personal life	Writing about time Management
Tonarelli et al. (2018)	3	20	Consecutive	Writing about thoughts and feelings about traumatic, stressful and emotionally significant event in personal life	Writing about time Management

Ten of the thirteen studies wrote on consecutive days and thus had no day off between the writing sessions. Participants in the study by Cosentino et al. (2019) wrote in a span of one to three days, and Cosentino et al. (2021) and Tarquini et al. (2016) had their participants write three times each in three weeks. In the instructions for the experimental group, five studies gave a clear invitation to reflect on occupational experiences. In the Ashley et al. (2013) study, this indication was made in an experimental condition. Six studies did not include this direct occupational reference in their instructions. Tarquini et al. (2016) presented both approaches. They asked their participants to reflect on their thoughts and feelings about the transfer (the reason for the intervention) and further provided the possibility to relate these thoughts and feelings to personal experiences. Nine control group instructions used a relatively trivial topic (e.g., time management, bedroom design, etc.). The remaining three studies asked participants to write about a significant event as well, but as objectively as they could (Cosentino et al., 2019; 2021; Procaccia et al., 2021).

Conclusion

Occupational stress is a problem that affects both employees and organizations. Therefore, the relevance of valid interventions to reduce occupational stress is high. Expressive writing is a well-structured written emotional disclosure intervention that has already been used effectively in an organizational context and is cost-effective and scalable. It is often characterized by its implementation protocol, which invites participants to write independently for 15–20 minutes, four to five times on different days, about their deepest thoughts and feelings regarding a stressful event.

This review examined thirteen studies that applied expressive writing in the workplace. The literature was identified using the PRISMA method (Moher et al., 2009). Studies were analyzed based on an assessment of internal validity, positive and unexpected effects, and differences in the protocol for implementing expressive writing.

Ten studies were randomized control trials, and three were quasi-experimental (see Table 2). Nine studies found a significant positive effect of expressive writing on psychological well-being or associated variables. The most notable improvements were in mental health variables (e.g., burnout, depressive symptoms, disturbed sleep) (Cosentino et al., 2021; Procaccia et al., 2021; Tarquini et al., 2016), personal resources (e.g., resilience, self-efficiency, self-acceptance) (Kirk et al., 2011; Saldanha & Barclay, 2021; Tarquini et al., 2016), as well as organizational variables (e.g., commitment, willingness to reconcile after an injustice, job satisfaction) (Barclay & Saldanha, 2016; Barclay & Skarlicki, 2009; Cosentino et al., 2021; Round et al., 2022). Unexpected effects occurred in four studies (Ashley et al., 2013; Cosentino et al., 2019; Michailidis & Cropley, 2019; Tonarelli et al., 2017). These findings will be discussed later in more detail.

Based on these results, the research question “How can expressive writing be used to address occupational stress?” can be answered as follows: The literature review provides evidence that expressive writing can reduce occupational stress. Occupational stress is a broad construct and should be distinguished in this discussion from specific stress concepts such as burnout syndrome. For example, the National Institute for Occupational Safety and Health (NIOSH) defines work stress as harmful physical and emotional reactions that occur when job demands do not match workers’ abilities, resources, or needs (Mohajan, 2012). On this basis, the JD-R Model can be used to justify the effectiveness of expressive writing and understood as a resource that increases a person’s coping potential. This additional coping potential has a buffering effect against demands and thus indirectly a reducing effect on the resulting harmful stress reaction. Whether this increase in coping potential is also reflected in reduced dimensions of burnout syndrome, such as increased exhaustion, increased cynicism, or reduced personal effectiveness (cf. Maslach et al., 1997) remains questionable. The effectiveness of the intervention for specific work stress concepts has yet to be identified. However, research shows positive effects on psychological health and organizational variables, suggesting at least an indirect effect.

Below, the results are discussed and structured based on the substantive alignment and coherence of the findings. This structure should reflect the antecendencies, consequences and limitations of the intervention and allows for a comprehensive exploration of the conditional factors, the impact on participants, and the potential constraints associated with the intervention. Antecedents refer to the conditional factors that should be considered when implementing expressive writing. Consequences refer to the effects of the intervention on the participants. Limitations, on the other hand, refer to the boundaries or constraints. These limitations are discussed in terms of factors that may restrict the effectiveness of expressive writing.

After discussing the content of the results, practical implications and limitations of this research are discussed.

A target group for expressive writing

Across different samples (clinical, non-clinical, occupational samples) expressive writing has recurring positive effects on psychological outcomes and health variables (Gao, 2022). However, these effects are not continuous. Repeated studies find no or ambiguous effects (e.g., Meads & Nouwen, 2005; Mogk et al., 2006; Reinhold et al., 2018). This pattern can also be seen in using the intervention in an organizational context. In meta-analyses and research on the use of the intervention in an organizational context, participant stress levels at baseline are discussed as a moderator of effectiveness (Ashley et al., 2013; Reinhold et al., 2018). Expressive writing particularly benefits people with higher stress levels (e.g., Cosentino et al., 2021; Procaccia et al., 2021). Individuals with sufficient personal resources to cope with stressful situations seem to benefit less (Kirk et al., 2011). Therefore, for the intervention to serve its purpose, expressive writing is suggested, especially with highly stressed employees. Highly stressed employees are defined as individuals who identify themselves as such through self-disclosure or those faced with stressful working conditions. These could be, for example, major changes in work life (e.g., restructuring, entering a leadership role) (see Tarquini et al., 2016) or an experience of injustice (see Saldhanda & Barclay, 2021). Expressive writing would be a cost- and time-efficient resource to provide to these high-stress employees to promote mental health and foster commitment (e.g., Cosentino et al., 2021; Procaccia et al., 2021).

Individuals with many resources and low-stress levels might also benefit from expressive writing, although the state of research here needs to be clarified. The risk of inefficient use of organizational and personal resources is minimal since Caputo et al. (2022), for instance, argue that expressive writing is justified by its lower cost even when the impact is low.

Opportunities and limitations of expressive writing in a professional context

The positive effect on mental health has been repeatedly demonstrated (e.g., Barclay & Skarlicki, 2009; Cosentino et al., 2021; Procaccia et al., 2021). The intervention positively influenced personal resources such as resilience, self-efficacy, and self-acceptance (Kirk et al., 2011; Saldanha & Barclay, 2021; Tarquini et al., 2016). This suggests that expressive writing reduces stress symptoms and offers preventive protection by building up personal resources. Expressive writing is therefore indicated as a psychological tool for occupational health management. Since organizational variables such as continuance commitment (Cosentino et al., 2021) and job satisfaction (Round et al., 2022; Tarquini et al., 2016) could be improved, it could be argued that expressive writing is also an effective intervention for human resource development. However, these effects are influenced by several aspects. The implementation protocol is to be mentioned here (Cosentino et al., 2019; 2021). It is necessary to define how many sessions are conducted, how the instructions are formulated, and in what order they are presented. A higher number of sessions with a longer duration seems to be more beneficial. The present study showed that positive effects occurred mainly when the number of sessions was between three and five sessions, with a duration of 20 minutes (see Tables 2 and 3). This finding is consistent with meta-analytic research that argues that session duration may explain differences in expressive writing effectiveness (e.g., Reinhold et al., 2018). In addition, Pennebaker and Chung (2011) also consider longer sessions conducted more frequently to be more effective. The premise is that relevant mental processes are mainly active between sessions and need time to develop.

This prerequisite of mental processing might be related to the unexpected results of Tonarelli et al. (2017). The authors hypothesized that reappraisal was not fully completed in their study due to the number of writing sessions (three sessions) and the duration of each session (20 minutes). However, they did not differ significantly from other studies regarding these parameters (see Table 3). Therefore, it is assumed that reappraisal is generally possible with three writing sessions of 20 minutes.

Another explanation for the unexpected results could be the wording of the instruction. Tonarelli et al. (2017) did not direct their participants to refer to a triggering situation at work and asked them to write about an emotionally upsetting and possibly even traumatic event. This wording may have led to the need for aftercare, which is not provided in the professional context. However, this argument is contradicted by the fact that participants in the 2018 study were also asked to deal with traumatic experiences. In this quasi-experimental study, the effect was not evident (Tonarelli et al., 2018). The impact of the wording of the instruction, nevertheless, should not be neglected. Other studies also show the influence of the instructions’ language on the intervention’s effectiveness (Barclay & Saldanha, 2016; Barclay & Skarlicki, 2009; Saldanha & Barclay, 2021). Research conducted by Barclay and Saldanha (Barclay & Saldanha, 2016; Saldanha & Barclay, 2021) showed positive effects of an additional instruction element that suggested incorporating sense-making or forgiveness. However, it is crucial that initial mental processing has already taken place, as their research from 2015 shows: Instructions with a forgiveness element only produced added value if they were formulated after the traditional structure of the intervention (write about thoughts and feelings) (Barclay & Saldanha, 2016). Therefore, it is recommended to formulate the instructions to relate to the professional context and invite intensive reflection on the cognitive and emotional aspects of an experience.

It should also be mentioned here that expressive writing is an intervention that affects different sociodemographic groups differently. In their meta-analysis, Reinhold et al. (2018) already discussed the possibility of influences of different person-related variables on the effectiveness of the intervention. For the organizational context, the study by Procaccia et al. (2021) supports this assumption. Their research shows that sociodemographic variables such as gender, age, and relationship status influence the effect of the intervention. Married participants seem to benefit more than unmarried participants. Lower age also appears to promote the impact of expressive writing. Regarding gender, women experience more social support after the intervention, while men experience greater improvement in terms of stress symptoms. The effectiveness of the intervention in relation to the demographic group should therefore be taken into account when it is used in the organization to avoid possible discrimination against certain groups.

Practical implications

The following protocol for using expressive writing in an organizational context can be derived from our conclusions. The frequency and duration of the writing task are critical for the effectiveness of the intervention and should not be less than four sessions of 20 minutes. Consecutive sessions are recommended based on the research presented. Instructions should be formulated to enable participants to engage intensively with their thoughts and feelings. References to a work-related topic or event are recommended. Terms such as "traumatic" should be avoided. In addition, aspects that can promote specific consequences of the intervention (e.g., forgiveness or sense-making) can be added. However, this should only be presented from the third session onwards to enable initial mental processes.

Limitations of this review and future research

The present study has some limitations that need to be considered. These are mainly related to the literature search and the literature review method.

Because of the interdisciplinary nature of the research question, it was decided to conduct the literature search in two discipline-independent databases (Scopus and Web of Science), one that focuses more on organizational management literature (Business Source Ultimate) and one that compiles psychological literature (PsycINFO). Nevertheless, relevant research may have been overlooked by the choice of databases and search terms. Therefore, no measure of comprehensiveness can be provided. In future research, it would be beneficial to incorporate terms related to the root word "work" (e.g., work, worker, etc.) into the search process. This could enhance the scope and comprehensiveness of the study.

Since a literature review was chosen as the method, the state of research in this paper is impaired, condensed, and syntagmatically presented. No data were collected, so no internal knowledge gain occurred. Also, there was only a very brief engagement with a theoretical foundation of the intervention grounded in theories and models of industrial and organizational psychology. Nevertheless, these limitations are more closely tied to the general methodological principle. Specifically pertaining to the current research, the evaluation and discussion strategy can be subjected to scrutiny based on antecedents, consequences, and limitations. It is worth noting that establishing clear boundaries and providing precise definitions for the terms involved can be challenging. Additionally, classifying the results based on antecedents, consequences, and limitations may be inadequate due to overlapping and ambiguous categorizations. Consequently, a more flexible content-based structure has been adopted for the discussion.

Future research could address these weaknesses and provide additional independent evidence, for example, through experimental studies. This could include further investigation of the moderators of intervention effectiveness. Preliminary evidence points to demographic variables, baseline stress levels, and the structure of the intervention. Organizational variables should also be included in future research, as this has not been done comprehensively so far. Also, the theoretical foundation could be deepened by examining whether, for example, the JD-R Model (Bakker & Demerouti, 2017) proves to be an appropriate theoretical foundation in the context of expressive writing.

Acknowledgement

Special thanks to Lena Dalkowski and Lara Meimouna Mbaye, student assistants at the German Police University in Münster, who contributed to the screening process as independent researchers.