When employees remain silent: explaining responses to suspected alcohol problems among co-workers

Introduction

Alcohol-related problems have substantial consequences for employees, organizations, and society. Alcohol use is a leading contributor to the global burden of disease and is associated with a wide range of health, social, and economic harms (World Health Organization, 2019; Rehm et al., 2017; GBD 2016 Alcohol Collaborators, 2018). In workplace settings, alcohol-related problems are associated with impaired work performance, sickness absence, workplace accidents, and productivity losses (Ames et al., 1997; Frone, 2013; Kirkham et al., 2015; Roche A. et al., 2016; Aas et al., 2017; Buvik et al., 2018).

Despite these consequences, alcohol problems often remain unidentified and untreated for many years (Chapman et al., 2015; Tuithof et al., 2016), allowing problems to escalate and increasing the risk of adverse health, social, and occupational outcomes.

An important reason for treatment delays is the stigma surrounding alcohol use and alcohol-related problems, both in society and in the workplace (Reynolds et al., 2008; Andréasson et al., 2013). Perceived stigma is associated with lower help-seeking and reduced willingness to disclose alcohol-related concerns (Wallhed Finn et al., 2023). Feelings of guilt and shame, as well as fear of negative reactions from others, have repeatedly been identified as barriers to disclosure and treatment seeking (Room, 2005; Schomerus et al., 2011). Consequently, many individuals delay seeking support for years or never seek treatment at all (Tuithof et al., 2016). Long delays between the onset of alcohol-related problems and help-seeking have been documented across countries, with individuals often waiting several years before accessing treatment (Chapman et al., 2015). Earlier identification is therefore critical to facilitate early intervention and prevent more severe consequences of the alcohol problem (Frone, 2013; Martinez et al., 2022; Morse et al., 2022).

The workplace represents a particularly important setting for early identification and prevention. Most adults spend a substantial proportion of their waking h at work, and employment rates remain high across Europe. In Sweden, approximately 5.2 million people were employed in 2025, corresponding to an employment rate of about 69% among individuals aged 15–74 years (EU labour market—Quarterly statistics, n.d.; Statistics Sweden, 2025). This proximity places managers and co-workers in a unique position to observe signs of developing alcohol-related problems. The preventive role of workplaces is further reinforced by legal and institutional frameworks. In Sweden, employers are required to prevent ill health and provide rehabilitation support, including for alcohol-related problems (Swedish Work Environment Authority, 2020; Swedish Work Environment Act, n.d.). Similar obligations exist across the European Union, Australia, and the United States, where workplace policies and support systems are increasingly viewed as important components of alcohol prevention and early intervention (Merrick et al., 2007; Pidd et al., 2011; Frone, 2013; Roche et al., 2015; Borrelli et al., 2022). Yet despite this preventive framework, substantial delays in identifying and addressing alcohol-related problems remain (Chapman et al., 2015). Recent qualitative research also suggests that workplaces play an important role not only in the early identification of alcohol-related problems but also in supporting employees during treatment and recovery, highlighting the broader role of organizations in responding to alcohol-related problems once they are recognized (Pistone et al., 2026).

Within this context, co-workers may play a particularly important role. Because employees often interact closely on a daily basis, they are frequently among the first to notice signs that a colleague may be struggling with alcohol-related problems. However, noticing a problem does not necessarily lead to action. Research on employee voice and silence suggests that individuals often refrain from raising sensitive concerns because they fear social, interpersonal, or professional consequences (Detert and Edmondson, 2011; Morrison, 2014). Morrison (2023) argues that silence is shaped by organizational norms, power relations, and perceptions of risk, and that speaking up is more likely when responsibilities are clear and psychological safety is high. Similarly, research on workplace bystander behavior shows that intervention is influenced by perceived personal risk and organizational climate, while fears of retaliation or social costs often inhibit action (Madden and Loh, 2020). Classic bystander theory further highlights how uncertainty and diffusion of responsibility can reduce intervention, particularly when others are also aware of the situation (Latané and Darley, 1968; Fischer et al., 2011; Hussain et al., 2019).

Research focusing specifically on alcohol-related concerns points to a similar gap between awareness and action. Studies have shown that alcohol use and intoxication occur in work-related contexts and may negatively affect co-workers and the work environment, yet relatively little is known about how employees respond when they suspect that a colleague has an alcohol-related problem (Frone, 2006; Moan et al., 2020). Earlier research found that approximately one-third of American psychologists were aware of colleagues whose alcohol misuse affected their work, yet only a minority reported confronting them (Thoreson et al., 1986). Alcohol-related concerns were addressed less frequently than mental health concerns, suggesting that alcohol problems may be perceived as particularly sensitive in workplace settings. More recently, Klingemann et al. (2026) found that although most healthcare professionals reported willingness to intervene in hypothetical situations, considerably fewer had acted when faced with real-life concerns. Supervisor support emerged as an important facilitator of action.

Emerging evidence further suggests that organizational conditions may influence whether employees translate concerns into action. Sundqvist et al. (2025) found that alcohol-related procedures combined with supervisor guidance were associated with greater comfort in reporting concerns, whereas the mere existence of an alcohol policy was not. Similarly, managers describe that organizational support structures, including HR support, education, policies, and visible leadership engagement, are essential for implementing alcohol prevention in practice (Sundqvist and Martinez, 2026). At the same time, employees may perceive alcohol-related concerns as outside their responsibility unless problems become severe (Tinghög, 2013). These findings suggest that both organizational support and individual perceptions of responsibility may shape whether concerns are voiced.

Substance use climate model

The present study draws on the Substance Use Climate Model (Ames and Grube, 1999; Frone, 2009, 2012), which emerged from earlier workplace alcohol research highlighting the importance of organizational culture, norms, and alcohol-related policies in shaping employee behavior (Ames et al., 1992). The model proposes that workplace environments shape alcohol-related attitudes and behaviors through both structural and social influences. Frone (2012) conceptualized substance use climate as comprising three interrelated dimensions: availability, descriptive norms, and injunctive norms. Availability refers to the extent to which alcohol is accessible in work-related settings; descriptive norms concern employees‘ perceptions of alcohol-related behaviors among colleagues; and injunctive norms reflect perceptions of the extent to which such behaviors are approved or disapproved of within the workplace. Although originally developed to explain employees' own substance use, the model can also be used as a framework for understanding how workplace contexts may influence employees‘ responses to colleagues' alcohol-related problems. Specifically, workplace norms and alcohol-related practices may shape whether employees perceive a concern as warranting action or whether they remain silent.

Psychological perspectives: employee voice, bystander behavior and stigma

Whereas the Substance Use Climate Model can be applied to the work environment, it does not account for the psychological or interpersonal processes through which an employee decides whether to act on a concern about a co-worker's alcohol consumption. The present study therefore also drew on psychological perspectives such as employee voice, bystander intervention, pluralistic ignorance, stigma, and health-related decision-making. These perspectives may help explain why recognizing a possible alcohol problem does not necessarily lead an employee to raise the issue.

Reasons for not voicing concerns may be related to the interpretation of the situation and problem ambiguity (Darley and Latané, 1968). Because alcohol-related problems develop gradually, an employee may be uncertain whether the observed behavior indicates a problem at all. If the colleagues show no sign of concern, their apparent unconcern may be taken as evidence that no problem exists, commonly referred to as pluralistic ignorance and social influence (Darley and Latané, 1968; Prentice and Miller, 1993).

Employees may also consider how serious the situation is and whether intervening would be helpful. Perspectives on health-related decision-making emphasize that action depends not only on recognizing a condition but also on determining whether it is serious enough to warrant response, and whether one believes that one's own action would be effective in helping the person (Health Belief Model; Rosenstock, 1974). Employees may therefore refrain from raising concerns when the co-worker appears to be functioning adequately or when there are no obvious consequences at work. They may also believe that raising the issue would make little difference, which is consistent with research on voice futility (Detert and Edmondson, 2011; Morrison, 2014).

Further explanations concerns whether intervening is regarded as one's own business. Responsibility may be attributed elsewhere when others are assumed to be aware of the situation (diffusion of responsibility; Latané and Nida, 1981; Darley and Latané, 1968), and stigma-related perspectives suggest that alcohol use is often construed as a private matter or as the individual's own responsibility, placing it outside the domain in which colleagues regard intervention as legitimate (Room, 2005; Schomerus et al., 2011).

Furthermore, silence is not merely the absence of voice but an active choice shaped by anticipated interpersonal and professional costs (Morrison, 2014, 2023). Employees may therefore remain silent where responsibility for acting is unclear or where raising a concern carries social risk, for instance when the person concerned holds a position of power. These theories informed the exploratory items on reasons for not voicing concerns in present study (see Table 1 below).

Table 1. Theoretical explanations for not voicing concerns.

This table represents outlines the theoretical foundations and references used to develop the exploratory items on why individuals may choose not to voice concerns.

Aims and research questions

Little is known about how employees respond when they suspect that a co-worker has an alcohol problem. Whereas, previous studies have largely examined hypothetical scenarios or general perceptions of reporting concerns, this study draws on employees‘ accounts of situations they had encountered in real life. The aim was to examine employees' responses to a co-worker's alcohol problem, and to identify organizational and individual factors associated with remaining silent. The following research questions were addressed:

• Is the availability of alcohol in work-related contexts associated with employees' likelihood of refraining from voicing concerns about a co-worker's alcohol use?
• Are descriptive norms regarding colleagues' alcohol-related behaviors associated with employees' likelihood of refraining from voicing concerns?
• Are attitudes toward alcohol-related behaviors associated with employees' likelihood of refraining from voicing concerns?
• What reasons do employees report for refraining from voicing concerns about a co-worker's alcohol use?

Materials and methods

This was a cross-sectional survey study. The data collection was conducted in December 2019 through a web-based questionnaire. The questions in the survey were designed to investigate whether Swedish employees had voiced concerns, the last time they suspected an alcohol problem at work. The survey also included other variables, such as work industry, employer, number of employees, alcohol availability and attitudes in relation to alcohol.

Participants and procedures

The participants were recruited from a Swedish citizen panel. The Swedish citizen panel was coordinated by “the SOM Institute,” which is an institute part of the University of Gothenburg, that focuses on research on society, opinion, and media habits among the Swedish population (The SOM Institute, 2026). For this study, an invitation was sent out to eligible members identified by The SOM Institute, which included adults (between the age of 18–80 years) who reported that they were employed. Eligible members for this study, who were randomly selected by the Citizen Panel, received a description of the study in which they were informed that the study was exclusively meant for adults who had a job. All studies conducted through this citizen panel have received ethical approval. The approval for this study was given during year 2018 (Exp. 2018-12-11, 1016-18) by the ethics committee the “Central Ethical Review Board” in Gothenburg. All participants in this study gave an informed consent regarding participation.

The survey was filled out by 3,302 participants. Participants were excluded for partial responses (responded to < 50 % of the questions), if employment was not the primary occupation (i.e., currently unemployed, self-employed, studying, retired). A group of 2,644 employed participants remained. Furthermore, participants (n = 901) were excluded if they reported never having suspected an alcohol problem at work, or if they did not respond to whether they had voiced their concerns or not. That left a final number of 1,743 participants. See Figure 1 for a flowchart and Table 2 presenting a detailed background description of the study sample. The data is divided into two groups: (1) participants who voiced their concerns and (2) those who refrained from voicing concerns when suspecting an alcohol problem.

Table 2. Background variables and study variables categorized by participants who voiced their concerns, and those who did not voice concerns.

This table presents the distribution of individual characteristics, work characteristics, alcohol preventive measures, alcohol availability, descriptive norms, and situational factors among employees who voiced concerns and those who refrained from voicing concerns about a co-worker's suspected alcohol problem. Total study sample N = 1,743. Valid N varies slightly across variables because of item-level missing data (0–0.6%). P-values are from Pearson chi-square tests.

Operationalisation of theoretical constructs concerning

The measures used in this study were selected to reflect the constructs described in the theoretical framework in the Introduction. Because the Substance Use Climate Model was developed to explain employees' own substance use, and because workplace alcohol practices in Sweden differ from those in the setting in which the model was developed, the dimensions required adaptation. Table 3 below summarizes each construct, their operationalisation, and their differences from the original model.

Table 3. Operationalisation of theoretical constructs related to substance use climate model.

This table represents the operationalisation of dimensions derived from the Substance Use Climate Model (Ames and Grube, 1999; Frone, 2009, 2012), including measurements, coding, and differences from the original model.

Workplace alcohol practices have changed substantially since the Substance Use Climate Model was first developed. In many Swedish workplaces, alcohol consumption during working his uncommon, whereas alcohol exposure is more likely to occur in work-related social contexts such as staff parties and after-work events. The construct availability was therefore operationalised as employer-provided alcohol at work-related occasions during the preceding 12 months, rather than as the perceived accessibility of alcohol during working h. This measure does not capture alcohol purchased by employees themselves or consumed at informal work-related events, and therefore likely underestimates total alcohol exposure in work-related settings.

Descriptive norms were operationalised as respondents‘ reported observations of three specific alcohol-related behaviors among colleagues. These items are intended to capture perceived descriptive norms as they arise from direct exposure to co-worker behavior within the substance use climate framework (Ames and Grube, 1999; Frone, 2012). Coding exposure as present vs. absent during the past 12 months follows prior applications of the substance use climate model, in which exposure to co-worker substance use is operationalised in precisely this way (Frone, 2012). The measure nonetheless differs from instruments that ask respondents to estimate the prevalence or frequency of such behavior among relevant others. It therefore captures whether alcohol-related behavior was observed, rather than respondents' quantitative estimates of how widespread it is.

Injunctive norms were not assessed. Instead, the survey item in this study intended to capture respondents' own appraisal of whether the behaviors described in a vignette indicated an alcohol problem, rather than perceived approval or disapproval among colleagues. This construct is therefore referred to as attitudes, and the findings are therefore not directly comparable with previous applications of the Substance Use Climate Model. The vignette was developed for the present study rather than adapted from a previously validated instrument. It was constructed to describe a situation in which problematic use of alcohol was observed and still considered ambiguous in interpretation, such as repeated hangovers during working hours and apparent intoxication at work-related social events. Further, it was specified that job performance remained adequate. This combination was chosen because it represents the kind of situation in which employees must determine about whether a concern is warranted, which was the interpretive process of interest in the present study. The vignette character was presented as either a woman (Anna) or a man (Adam), with participants randomly assigned to one version.

The exploratory items concerning reasons for not voicing concerns were informed by theoretical perspectives on employee voice, bystander behavior, voice futility, pluralistic ignorance and social influence, stigma, and situation appraisal described in the introduction. Table 1 below presents different theories and key references used to develop the exploratory items on why individuals may choose not to voice concerns.

Measures

Dependent variable

Not voicing concerns

In the survey, employees who suspected that a co-worker might have an alcohol problem were asked to respond to the following question: “Did you bring up your suspicions with anyone in the workplace?” The response options were: “Yes, with the person whom I suspected had an alcohol problem,” “Yes, with one or multiple co-workers (who were not part of Human Resources or did not have staff responsibility),” “Yes, with my manager,” “Yes, with another manager,” “Yes, with Human Resources or a person with staff responsibilities.” In contrast, the response option “No, with none of the above” was interpreted as “not voicing the concerns.” This variable was subsequently dichotomized into two categories:

• “Did not voice one's concerns” (1): Refers to cases where the participant did not bring up their suspicions with anyone in the workplace regarding their co-worker's potential alcohol problem.
• “Voice one's concerns” (0): Refers to cases where participants brought up their suspicions with someone in the workplace (i.e., with at least one of the aforementioned response options) the last time they suspected that a co-worker had an alcohol problem.

Interpretation of the outcome variable

The dependent variable identifies whether the concern was raised with anyone in the workplace. In the present study, “not voicing concerns” refers only to the observable response of not discussing the suspicion with anyone in the workplace. Employees may refrain from speaking up for different reasons, and such potential explanations were therefore examined separately in the exploratory items described further below.

Independent variables

Availability

Availability was operationalised as the extent to which alcohol was provided by the employer in work-related contexts. Participants were asked to report how many times their employer had provided alcohol during work-related occasions over the previous 12 months. The original response options were: Never (0), Once (1), 2–5 times (2), 6–10 times (3), 11–20 times (4), 21–40 times (5), and More than 40 times (6). Due to low frequencies in the higher response categories, these were collapsed into four categories: Never (0), Once (1), 2–5 times (2), and 6 times or more (3).This measure captures employer-provided alcohol in work-related settings but does not include alcohol that may have been purchased or provided by employees themselves or consumed in informal work-related social contexts.

Descriptive norms

Descriptive norms were operationalised as employees' reported observations of alcohol-related behaviors among others in work-related contexts. Participants were asked whether they had experienced that someone in their workplace (employee, colleague, or manager) had engaged in specific alcohol-related behaviors (“Have you experienced that someone in your workplace (employee, colleague, or manager) has…”).

The items included: “Appeared intoxicated at a staff party or similar event” (1), “Smelled of alcohol during working h” (2), and “Been hungover during working hours” (3).

The response options were: “Yes, during the past week” (1), “Yes, during the past month” (2), “Yes, during the past year” (3), “Yes, during the past 3 years” (4), “Yes, more than 3 years ago” (5), and “No” (6). For statistical analyses, responses were dichotomized into “Yes, within the past 12 months” (including past week, past month, and past year) vs. “No or more than 12 months ago” (including past 3 years, more than 3 years ago, and no).

This dichotomization approach is consistent with prior operationalizations of descriptive norms within the substance use climate framework (Frone, 2012), in which exposure to co-worker substance use is coded as the presence vs. absence of any exposure during the past 12 months.

Attitudes

Attitudes were assessed using a vignette in which respondents evaluated whether specific alcohol-related behaviors indicated an alcohol problem. The scenario described a co-worker who repeatedly showed up to work hungover and appeared intoxicated at work-related social events, while maintaining adequate job performance. Respondents were then asked to rate how likely they believed it was that the co-worker had an alcohol problem.

The response options were: “Very likely” (1), “Somewhat likely” (2), “Not very likely” (3), and “Not at all likely” (4). Due to the small number of responses in the “Not at all likely” category (n = 25), this category was combined with “Not very likely,” resulting in three categories. The construction of the vignette, and the relationship between this measure and the injunctive-norms dimension of the Substance Use Climate Model, are described above (see Operationalisation of theoretical constructs).

Covariates

Gender

The response options for gender included woman, man and non-binary. Since none of the participants responded being non-binary, the gender variable was dichotomized into man and woman.

Education

The response options for education were compulsory school, upper secondary school, post-secondary school and college/university. Since about half of the participants reported having a college/university degree, the variable for education was dichotomized into “college/university degree” vs. “no college/university degree”.

Age

The variable “age” (in years) included the categories “18–39”, “40–49”, “50–59” and “above 60”.

Time since suspicion

Time since the most recent suspicion of a co-worker having an alcohol problem was dichotomised as ≤ 5 years ago and >5 years ago.

Employer

The response options for the variable “employer” included governmental, municipal, county council/regional, private, non-profit organizations/foundations and other.

Number of employees

The variable number of employees was divided as followed: 1–9, 10–49, 50–249, 250–499, 500–1000 and more than 1000.

Manager position. Manager position included the two categories: “manager” or “not manager”.

Staff responsibility

Staff responsibility included the two categories: “having staff responsibility” or “having no staff responsibility.”

Alcohol preventive measures

Employees' awareness of the presence of workplace alcohol-preventive measures was assessed using four items: (1) the existence of an alcohol policy, (2) rules concerning occasions on which alcohol consumption was permitted at work or in work-related contexts, (3) alcohol-related education, and (4) standard procedures for responding to an existing alcohol problem in the workplace. Response options were “Yes”, “No” and “I don't know” and each item was dichotomised into “Yes”, vs. “No or don't know”. The four individual measures are presented separately in the descriptive analyses. For the regression analyses, a binary variable was constructed distinguishing respondents who reported at least one of these measures from those who reported none.

Exploratory questions

We explored different potential reasons for why individuals chose not to discuss their concerns about their co-worker's potential alcohol problem with someone in the workplace. In exploring this, we investigated 13 statements (potential reasons) that were asked to the individuals who reported that they refrained from voicing concerns. The response options for each statement were the following: “Does not apply at all” (1), “Does not apply very well” (2), “Applies somewhat” (3), and “Applies precisely” (4). The items were based on the psychological theories described above. Because participants rated each statement independently, the mechanisms were not mutually exclusive, participants could endorse several different reasons for refraining to voice concerns. The items were therefore analyzed separately rather than combined into scale scores, consistent with the exploratory nature of the research question.

The statements are reproduced below using their original questionnaire wording. In this context, “the person” referred to the co-worker whom the participant suspected of having an alcohol problem.

“I did not voice my concerns because…”

• The person with the (potential) alcohol problem was performing his/her work duties adequately (Rosenstock, 1974).
• I was not certain that the person had an alcohol problem (Darley and Latané, 1968).
• I was not (sufficiently) concerned about the person (Rosenstock, 1974).
• The person's (potential) alcohol problem did not cause any negative consequences in the workplace (Rosenstock, 1974).
• There were other people in the workplace (besides myself) who were also aware of the person's (potential) alcohol problem (Darley and Latané, 1968; Latané and Nida, 1981).
• I did not consider the person's (potential) alcohol problem to be particularly serious (Rosenstock, 1974).
• I believed that my actions would not make a difference for the person with the alcohol problem (Detert and Edmondson, 2011; Morrison, 2014).
• I considered the person's drinking habits to be a private matter (Room, 2005; Schomerus et al., 2011).
• I did not dare to bring up my concerns (Morrison, 2014, 2023).
• I believed that the alcohol problem was the person's own responsibility (Room, 2005; Schomerus et al., 2011).
• I feared that the person with the alcohol problem might face trouble or negative consequences if I voiced my concerns (Morrison, 2014, 2023).
• No one else in the workplace besides myself seemed to suspect that the person had an alcohol problem (Darley and Latané, 1968; Prentice and Miller, 1993).
• I feared that I might face trouble or negative consequences if I said something (Morrison, 2014, 2023).

Statistics

Descriptive statistics were calculated to summarize sample characteristics and study variables. Differences between participants who voiced concerns and those who did not were examined using Pearson's chi-square tests for categorical variables, as appropriate.

To examine factors associated with refraining from voicing concerns, binary logistic regression analyses were conducted. A model including the main independent variables without covariate adjustment was first estimated, followed by a full model in which the independent variables and covariates were entered simultaneously. The dependent variable was dichotomous, coded as 1 = did not voice concerns and 0 = voiced concerns. Odds ratios (ORs) with 95% confidence intervals (CIs) were estimated.

The primary model included variables derived from the Substance use climate model, including alcohol availability, descriptive norms, and attitudes. Further, relevant covariates (gender, age, education, managerial position, time since suspicion, whether the suspicion concerned a manager, and alcohol preventive measures at the workplace) were included. All independent variables were entered simultaneously using the enter method, allowing for estimation of adjusted associations.

Categorical predictors were included as dummy variables, with reference categories defined a priori (e.g., lowest exposure or theoretically meaningful baseline categories). For ordinal variables (e.g., alcohol availability and age), categories were retained rather than treated as continuous to avoid assumptions of linearity.

Model fit was evaluated using several complementary indicators. The Omnibus test of model coefficients was used to assess overall model significance. Nagelkerke's R² is reported as a descriptive index. Because pseudo-R² statistics in logistic regression do not represent the proportion of variance explained in the sense familiar from linear regression, and are sensitive to the base rate of the outcome and to the number and type of predictors included, this value is not used as a criterion of model adequacy. Calibration of the models was assessed using the Hosmer–Lemeshow goodness-of-fit test, where a non-significant result indicates acceptable fit. Classification tables were examined to assess the proportion of correctly classified cases.

To assess the robustness of the findings and to distinguish between different forms of speaking up, two supplementary analyses were conducted using logistic regression. In the first supplementary analysis, the outcome compared participants who reported their concerns to a co-worker with formal organizational responsibility (e.g., a manager or HR) with those who did not speak to anyone, thereby excluding participants who spoke only to colleagues without staff responsibility. In the second analysis, participants who only raised concerns with colleagues without staff responsibility were compared with those who took a more direct or formal form of action, defined as either approaching the suspected co-worker directly or raising the concern with someone with formal organizational responsibility. Direct communication with the suspected co-worker was included in this latter category even though it is not a formal organizational reporting. The reason for this inclusion was that it was still considered a more direct intervention compared to merely discussing the concern with a colleague without any staff responsibilities. Thus, the second analysis should be understood as contrasting informal third-party discussion with direct interpersonal or formal organizational responses, rather than as a comparison of informal vs. formal reporting.

Missing data were limited across the main study variables. Among the variables included in the descriptive analyses, item-level missingness ranged from 0% to 0.6%. Regression analyses were conducted using complete cases (listwise deletion). The analytical sample was N = 1,727 for the model including the main independent variables and N = 1,720 for the fully adjusted model. The two supplementary analyses were based on analytical samples of N = 1,286 and N = 1,315, respectively. The reduced sample sizes in the supplementary analyses primarily reflect the restriction of these analyses to the response groups being compared, rather than item-level missing data. All tests were two-tailed, and a p-value < 0.05 was considered statistically significant. All statistical analyses were conducted using IBM SPSS Statistics.

Results

Out of all participants, 403 (23.1%) refrained from voicing their concerns, while 1,340 (76.9%) reported that they had voiced their concerns. Binary logistic regression models were estimated to examine factors associated with refraining from voicing concerns about a co-worker's suspected alcohol problem. The models included indicators of descriptive norms, alcohol availability, attitudes, and relevant covariates. A main independent-variable logistic regression model was first performed, followed by a full model including all predictors and covariates.

Main independent-variable binary logistic regression model

The main independent-variable model estimated the associations between the independent variables (availability, descriptive norms, and attitudes) and refraining from voicing concerns. In this model, higher alcohol availability (six times or more compared to never) was associated with a higher likelihood of not voicing concerns [OR = 2.10, 95% CI (1.25–3.54), p = 0.005]. In addition, employees who perceived the scenario as very likely to indicate an alcohol problem were less likely to refrain from voicing concerns compared to those who considered it not likely (OR = 0.59, 95% CI (0.42–0.83), p = 0.003). Having observed a co-worker intoxicated at a staff party was associated with a lower likelihood of refraining from voicing concerns (OR = 0.75, 95% CI [0.57–0.99], p =0.044).

Binary logistic regression (full model)

A full model including all predictors and covariates was estimated. The model was statistically significant, χ²(17) = 87.61, p < 0.001. Nagelkerke's R² was.075. Calibration was acceptable (Hosmer–Lemeshow test: χ²(8) = 12.64, p =0.125). See Table 4 for an overview of the full model displaying factors associated with refraining from voicing concerns.

Table 4. Binary logistic regression models of factors associated with refraining from voicing concerns.

This table presents the results from the main independent-variable and fully adjusted binary logistic regression models examining factors associated with refraining from voicing concerns about a co-worker's suspected alcohol problem. The dependent variable in the binary logistic regression models was “refraining from voicing concerns.” Initially, a model including the main independent variables (availability, attitudes, and descriptive norms) was estimated without covariates. In the fully adjusted model, the independent variables were tested together with the covariates. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Statistically significant associations (p < 0.05) are shown in bold.

Descriptive norms

Among the descriptive norm variables, having observed a co-worker intoxicated at a staff party within the past year was significantly associated with a lower likelihood of not voicing concerns [OR = 0.75, 95% CI (0.57, 0.99), p = 0.044]. In other words, employees who had observed intoxication at a staff party were more likely to speak up about their concerns compared to those who had not observed such behavior. In contrast, having experienced that a co-worker smelled of alcohol during working h or had been hungover was not significantly associated with voicing behavior (p > 0.05).

Availability

Alcohol availability was measured as how often the employer provided alcohol in work-related contexts. The results showed that employees who reported that alcohol was provided six times or more during the past year had significantly higher odds of not voicing concerns [OR = 2.00, 95% CI (1.16, 3.45), p = 0.013] compared to those who reported that alcohol was never provided.

Attitudes

The overall association between attitudes and refraining from voicing concerns was not statistically significant (p =0.090). However, the contrast between participants who rated the situation as “very likely” to indicate an alcohol problem and those who rated it as “not likely” was statistically significant. Participants in the “very likely” category had lower odds of refraining from voicing concerns (OR = 0.69, 95% CI [0.48, 0.98], p = 0.038).

Alcohol preventive measures

The presence of at least one reported alcohol-preventive measure was associated with a lower likelihood of refraining from voicing concerns [OR = 0.84, 95% CI (0.75, 0.95), p = 0.003].

Covariates

Several demographic and situational factors were significantly associated with voicing concerns. Women were less likely than men to refrain from voicing concerns (OR = 0.70, p = 0.005). Older employees, particularly those above 60 years, and those in managerial roles were also less likely to refrain from voicing concerns (OR = 0.45, p < 0.001; OR = 0.71, p = 0.022). In contrast, when the suspected individual was a manager, employees were more likely to refrain from voicing concern (OR = 1.55, p =0.014). Longer time since the suspicion occurred was associated with a higher likelihood of refraining from voicing concerns (OR = 1.63, p < 0.001).

Because a substantial proportion of participants who voiced concerns reported that they only spoke with colleagues without staff responsibilities (33.2%), additional supplementary analyses were conducted to distinguish between different forms of responses to suspected alcohol problems, comparing informal discussion with colleagues with more direct or formal responses, defined as either approaching the suspected co-worker directly or reporting the concern to someone with formal organizational responsibility. These analyses aimed to further examine the robustness of the main findings and to clarify whether the factors associated with voicing concerns differed depending on whom employees spoke to.

The first supplementary analysis (Supplementary Table 1) compared participants who reported their concerns to someone with formal responsibility (e.g., a manager or HR) with participants who did not voice concerns to anyone in the workplace. Participants who only spoke with colleagues without staff responsibilities were excluded from this analysis. The model was statistically significant (χ²(17) = 134.89, p < 0.001) and demonstrated good calibration (Hosmer–Lemeshow p = 0.216), with a Nagelkerke R² of.140. Because this model was estimated on a restricted sample with a different outcome base rate, its pseudo-R² is not directly comparable with that of the main analysis, and the difference should not be read as evidence that the predictors perform better in an absolute sense.

The supplementary analyses revealed both similarities and differences compared with the main analysis, suggesting that associations depended partly on the type of employee response being tested. In the first analysis (Supplementary Table 1), which compared formal reporting with not voicing concerns, several associations observed in the main analysis were retained, including those for attitudes, alcohol-preventive measures, age, managerial position, and whether the suspicion concerned a manager. However, some associations differed. Observing a co-worker intoxicated at a staff party was not significantly associated with formal reporting vs. silence [OR = 0.78, 95% CI (0.58, 1.06), p = 0.111], and the estimate for alcohol availability of six or more occasions was attenuated and no longer statistically significant [OR = 1.66, 95% CI (0.92, 2.99), p = 0.093]. In contrast, alcohol availability on 2–5 occasions was associated with the outcome [OR = 1.50, 95% CI (1.08, 2.09), p = 0.016].

The second analysis (Supplementary Table 2) excluded participants who reported that they did not bring up their concerns to anyone in the workplace. Instead, this analysis compared participants who voiced concerns only to colleagues without staff responsibilities with those who reported concerns to someone with formal responsibility or directly to the co-worker with the alcohol problem. The model was statistically significant (χ²(17) = 126.72, p < 0.001) and had a Nagelkerke R² of 0.128, which for the same reason is not directly comparable with that of the main model. That model also had a weaker fit (Hosmer–Lemeshow p = 0.038). In this model, high alcohol availability (six or more occasions during the past year) was associated with a lower likelihood of voicing concerns only to colleagues (OR = 0.47, p = 0.045).When employees who spoke only to colleagues without staff responsibility were compared with those who reported concerns to someone with formal responsibility or directly to the suspected co-worker, observing intoxication at a staff party was essentially unrelated to the outcome [OR = 0.98, 95% CI (0.74, 1.31), p = 0.908]. Moreover, alcohol availability of six or more occasions showed an association in the opposite direction [OR = 0.47, 95% CI (0.22, 0.98), p = 0.045]. The supplementary analyses showed that the pattern of associations differed depending on the type of employee response being compared.

Exploratory question

We explored the reasons why participants refrained from voicing concerns (Figure 2). The most reported (71.4%) reason was the belief that the co-worker with the alcohol problem was performing their job duties adequately.

The second most frequently reported reason (71.3%) was uncertainty about whether the co-worker actually had an alcohol problem. The third most common reported reason (69.5 %) was insufficient concern for the co-worker. The fourth most cited reason (65.5 %) was that the co-worker's alcohol problem did not result in any negative consequences in the workplace. The fifth most frequently reported reason (60.3%) was that other employees in the workplace were also aware of the situation.

Discussion

Summary and interpretation of findings

The study examined factors associated with employees refraining from voicing concerns when they suspected that a co-worker had an alcohol problem. Approximately one in four employees reported remaining silent, highlighting a gap between recognizing a potential problem and taking action. Overall, the findings suggest that both workplace alcohol climate and individual perceptions contribute to employees' responses, although a substantial share of the variation in employees' responses was not accounted for by the measures available in this survey. This may indicate that other psychological and interpersonal processes are important for understanding why employees remain silent. Based on the exploratory findings, uncertainty about whether a problem exists and perceptions of the seriousness of the alcohol problem may predict whether employees decide to intervene (Rosenstock, 1974). Furthermore, perceptions of diffused responsibility because others are assumed to be aware of the situation may also inhibit employees from vocing concerns (Darley and Latané, 1968; Latané and Nida, 1981). In addition, anticipated interpersonal or professional consequences may discourage employees from speaking up (Detert and Edmondson, 2011; Morrison, 2014, 2023). These processes may complement the Substance Use Climate Model in explaining employees' responses to suspected alcohol problems among co-workers.

The findings in this study still provide partial support for the Substance Use Climate Model (Ames and Grube, 1999; Frone, 2009, 2012). Employees working in environments where alcohol was more frequently provided by the employer in work-related contexts were more likely to refrain from voicing concerns. This is consistent with previous research suggesting that permissive workplace alcohol climates may normalize alcohol-related behaviors and reduce the perceived need to intervene (Frone, 2012, 2013). In such contexts, behaviors that might otherwise trigger concern may be interpreted as socially acceptable or less problematic.

The modest effect sizes in this study further suggest that recognizing a an alcohol problem is only one step in the process and does not necessarily lead to action (Sheeran and Webb, 2016).

This interpretation is also supported by the exploratory findings, which indicated that employees who remained silent frequently reported that the co-worker was still performing their work adequately, that they were uncertain whether an alcohol problem actually existed, that they were not sufficiently concerned, that there were no negative consequences in the workplace, or that other employees were already aware of the situation. These reasons suggest that remaining silent may reflect several different processes. Employees may first need to recognize a potential problem, interpret it as sufficiently serious to warrant action, and perceive some personal responsibility for responding before acting on their concerns. When a situation is ambiguous or when others are believed to be aware of it, responsibility may instead be diffused, reducing the likelihood of intervention (Darley and Latané, 1968; Latané and Nida, 1981).

The presence of at least one alcohol preventive measures at work was also associated with a lower likelihood of remaining silent. This finding supports previous research suggesting that preventive measures such as alcohol policies, alcohol education, and clear procedures may facilitate employee voice and intervention behavior (Sundqvist et al., 2025; Martinez et al., 2022). Such organizational measures may be important not only because they provide practical guidance but also because they can clarify that alcohol-related problems are legitimate workplace concerns and to whom concerns should be reported. This may be particularly relevant in situations where employees are uncertain about whether or how they should act.

The supplementary analyses further suggest that employee responses to suspected alcohol problems should not be conceptualized simply as a binary distinction between speaking up and remaining silent in future studies. For example, the association with observing intoxication at a staff party was present in the primary model but was not evident when formal reporting was compared with employee silence, or when informal responses were compared with formal reporting. Further, alcohol availability was associated with employee silence in the primary analysis and was attenuated in the formal-reporting-vs. silence analysis. However, the association was reversed in the comparison of informal with formal or direct responses. These findings suggest that workplace alcohol climate may influence not only whether employees respond to a suspected problem, but also how they respond. This heterogeneity is theoretically plausible because different forms of voicing involve different interpersonal and organizational demands. Discussing a concern informally with a colleague may require relatively little perceived authority or organizational support, whereas approaching the suspected co-worker or reporting to a manager or HR may involve greater perceived social risk (Morrison, 2014). Further, different psychological mechanisms may be involved in different forms of reporting. The higher pseudo-R² observed in the first supplementary analysis should not be interpreted as indicating better predictive performance than the primary model because the models were estimated using different samples and outcome definitions. The supplementary findings therefore extend the primary analysis by indicating that employee voice in relation to alcohol-related concerns may comprise several behavioral responses rather than a single continuum from silence to speaking up (Detert and Edmondson, 2011; Kish-Gephart et al., 2009; Morrison, 2014). Future research should examine these separate responses in more detail and accuracy, including the organizational and interpersonal conditions that influence employees' choice of whom to approach.

Theoretical implications

The study contributes to the workplace alcohol literature by extending the Substance Use Climate Model beyond its original purpose. The model was originally developed to explain employees' own alcohol and substance use behaviors in relation to workplace availability and norms (Ames and Grube, 1999; Frone, 2012). In the present study, however, the model was applied to a different outcome: employees' responses to suspected alcohol problems among co-workers. The findings suggest that workplace alcohol climate may be associated not only with employees' own substance use but also with their responses when concerns about a co-worker arise. At the same time, the results indicate that substance use climate dimensions capture only part of the mechanisms underlying employee responses. The findings therefore suggest that workplace alcohol climate should be integrated with broader theoretical perspectives from employee voice research and psychological theories. Further, the increased likelihood of remaining silent when the suspected individual was a manager further highlights the importance of power relations and perceived social risks; themes that are central to contemporary employee voice research (Morrison, 2023). The findings suggest that employees' responses to co-workers' alcohol problems are associated with several organizational, interpersonal, and situational factors, including workplace norms, organizational support, interpersonal relationships, and characteristics of the situation.

Practical implications

The findings suggest that workplace alcohol prevention efforts should move beyond the existence of formal policies and address the factors that prevent employees from voicing concerns (Martinez et al., 2022). Employees frequently reported uncertainty about whether a co-worker had an alcohol problem and often perceived the individual as functioning adequately at work. Organizations may therefore benefit from providing clearer guidance on early warning signs and emphasizing that concerns can be raised before work performance is noticeably affected.

The association between alcohol preventive measures and a lower likelihood of employee silence highlights the importance of organizational support structures. Clear procedures, designated points of contact, and visible organizational support may reduce uncertainty about how concerns should be handled. Employees should also be informed about available reporting responses and when concerns should be raised with managers, HR, or other responsible actors.

The exploratory findings further suggest that workplace interventions should address bystander processes. Many employees assumed that others were already aware of the alcohol problem, or questioned whether their actions would make a difference. Training and communication efforts may therefore benefit from emphasizing individual responsibility and encouraging employees to act on concerns rather than assuming that someone else will intervene.

Finally, although reducing alcohol availability in work-related contexts may contribute to a less permissive workplace alcohol climate (Ames and Grube, 1999; Frone, 2012), the findings suggest that structural measures alone are unlikely to be sufficient. Effective prevention is likely to require a combination of organizational support, clear reporting procedures, and workplace norms that encourage employees to raise concerns when needed (Detert and Edmondson, 2011; Morrison, 2014, 2023; Roche et al., 2015; Sundqvist et al., 2025).

Potential factors influencing employee silence

The statistical models accounted for only part of the variation in whether employees voiced their concerns, suggesting that other relevant factors were not captured by the available measures. The exploratory items provide some clues about factors that may be relevant. Participants who remained silent frequently reported reasons relating to uncertainty about the situation, the assumption that others were already aware of it, doubts about whether their own actions would make any difference, and the view that a colleague's drinking was a private matter. These questions were asked retrospectively and only to participants who had refrained from speaking up, so they cannot be examined as predictors in the present study. However, they highlight several factors that could be assessed directly in future research.

Psychological safety, the shared belief that a work group is safe for interpersonal risk-taking, is among the most consistently documented antecedents of employee voice and can be assessed with the seven-item Team Psychological Safety Scale (Edmondson, 1999), although doing so requires respondents to be nested within identifiable work groups rather than sampled individually, as in the present study. Psychological safety may be particularly relevant in this context because raising concerns about a colleague's drinking can involve considerable interpersonal risk. Employees' expectations about what might happen if they speak up may also be important. Previous research suggests that both perceived risk and expectations that speaking up will make little difference can contribute to employee silence (Detert and Burris, 2007; Detert and Edmondson, 2011; Kish-Gephart et al., 2009). Measuring both is particularly relevant given the present finding that employees were more likely to remain silent when the co-worker that was suspected was a manager, a situation in which anticipated costs and doubts about whether speaking up would change anything are plausible.

Factors at the organizational level may also influence whether employees raise concerns. Vertical trust and organizational justice are available in validated Swedish-language form in the Copenhagen Psychosocial Questionnaire (COPSOQ III; Berthelsen et al., 2020), and perceived organizational support can be assessed with the short form of the Survey of Perceived Organizational Support (Eisenberger et al., 1986); together these would allow future studies to test whether the association observed here between alcohol-preventive measures and speaking up reflects those measures.

Finally, 60.3% of those who remained silent reported that others in the workplace were also aware of the situation, which is consistent with diffusion of responsibility. The survey did not, however, assess whether these employees regarded intervening as part of their own role. Future studies investigating employee silence in relation to alcohol in work contexts could potentially adapt the five-item self-report measure of felt responsibility for constructive change (Morrison and Phelps, 1999), so as to capture whether employees regard responding to a colleague's suspected alcohol problem as falling within their own role rather than someone else's.

Limitations and future research

The cross-sectional design precludes conclusions regarding causality or the direction of the observed associations. In addition, all measures relied on self-report, making the results susceptible to recall bias and social desirability bias. Further, some key constructs were operationalised differently from the original Substance Use Climate Model. Most notably, the study measured respondents‘ own evaluations of alcohol-related behaviors rather than injunctive norms, which traditionally refer to perceptions of co-workers' approval or disapproval of such behaviors (Frone, 2012). Consequently, direct comparisons with previous applications of this model should be made cautiously. Similarly, descriptive norms were assessed through reported observation of specific behaviors rather than through estimates of their prevalence among colleagues, which may capture normative perceptions less directly than prevalence-based measures.

Moreover, alcohol availability was operationalised as employer-provided alcohol in work-related contexts. This measure does not capture other forms of alcohol availability that may be relevant to workplace alcohol climate, such as alcohol purchased by employees themselves, provided by unions, or consumed during informal work-related social events. The measure therefore likely underestimates actual alcohol exposure in work-related settings.

Another limitation is that the measure of workplace alcohol prevention also represents a relatively broad operationalisation. Although the four individual measures (policy, rules concerning alcohol use, education, and procedures for responding to alcohol-related problems) were reported separately in the descriptive analyses, they were combined into a binary indicator of whether at least one measure was present in the regression analyses. These measures represent conceptually distinct organizational practices and may differ in their relevance to employees‘ willingness to voice concerns. The aggregated measure therefore cannot determine which specific preventive measures, or combinations of measures, account for the observed association. The finding should consequently be interpreted as reflecting employees' reported awareness of at least one workplace alcohol-preventive measure rather than the effect of any intervention.

Furthermore, the study sample was intentionally broad and included employees from a wide range of industries, sectors, organizational sizes, and workplace contexts across Sweden. Although this diversity strengthens the generalizability of the findings, it may also have contributed to the substantial unexplained variation in the regression models, since workplace-specific contextual factors of likely relevance to voice were not measured (see “Potential factors influencing employee silence”). An additional limitation concerns the heterogeneity of the primary outcome. Although not voicing concerns was operationalised as a single behavioral outcome, the exploratory findings indicate that this behavior may arise from several theoretically distinct mechanisms, including problem ambiguity (Darley and Latané, 1968), perceptions of severity and efficacy (Rosenstock, 1974), diffusion of responsibility (Latané and Nida, 1981), privacy and responsibility attributions (Room, 2005; Schomerus et al., 2011), and anticipated costs (Morrison, 2014, 2023). The primary regression analyses do not distinguish between these underlying mechanisms. The binary outcome therefore indicates whether participants raised their concerns, but it does not identify the reasons why some participants remained silent. Future studies should assess these mechanisms as distinct constructs and examine whether they have different organizational and individual antecedents.

Finally, the supplementary analyses suggest that different forms of employee response may represent partly distinct behavioral processes. Future studies should therefore distinguish between different types of intervention behavior and investigate the situational, relational, and organizational factors that influence each form of response.

Conclusion

Employees are often among the first to notice signs that a co-worker may have alcohol-related problems, yet recognizing a concern does not necessarily lead to action. In this study, nearly one in four employees who suspected that their co-worker had an alcohol problem refrained from voicing concerns. Employees were more likely to remain silent in workplaces where alcohol was more available in work-related contexts. However, the findings suggest that employee silence surrounding suspected alcohol problems is shaped not only by individual judgements but also by the broader organizational context in which concerns arise. Uncertainty about whether a problem exists, perceptions that the individual is functioning adequately, and assumptions that others are already aware of the situation emerged as common reasons for remaining silent. These findings resonate with theories of employee voice and bystander intervention, highlighting how ambiguity and diffusion of responsibility may inhibit action even when concerns are recognized. Although a substantial share of the variation in employees' responses remains to be explained, the results highlight the importance of creating workplace environments that facilitate early intervention. Organizations may benefit from providing clear guidance on recognizing and responding to alcohol-related concerns and establishing visible reporting pathways. By reducing uncertainty and clarifying responsibility, workplaces may be better positioned to identify alcohol-related problems at an earlier stage and support employees before difficulties escalate. Ultimately, promoting a climate in which concerns can be raised safely and appropriately may represent an important, yet underutilized, component of workplace alcohol prevention.

Ethics approval and consent to participate

All studies conducted through the SOM Institute have received an ethics approval. The approval for this study was given during year 2018 (Exp. 2018-12-11, 1016-18) by the ethics committee the “Central Ethical Review Board” in Gothenburg. All participants in this study gave an informed consent regarding participation.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/forgp.2026.1938613/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Francesco Marcatto, University of Trieste, Italy

Reviewed by: Antonio Brusini, Local Health Unit of Modena, Italy

Federica Silveri, University of Studies G. d'Annunzio Chieti and Pescara, Italy