Commentary on Tajima et al. (2024): Interplay between disaster‐related experiences, stressors, and problem drinking

Following the Fukushima Daiichi nuclear power plant disaster in 2011, a large longitudinal study was conducted to examine the interplay between disaster exposure and problem alcohol use among a unique cohort of workers from the Daiichi and Daini plants. The study (Tajima et al., 2024), which followed participants over a three-year period following the disaster, provides insights into long-term impacts of disaster-related experiences and stressors on individuals' drinking behaviors. Tajima et al. (2024) found that there was an increase in the prevalence of problem drinking in the first 3 years after the disaster. Furthermore, disaster exposure and related experiences were associated with the pattern and longitudinal course of problem drinking, with some differences across different disaster-related experiences. Overall, these findings parallel previous literature that disaster-related experiences are associated with heightened risk for alcohol and substance use and related problems and mental health challenges (Fergusson et al., 2014; Flory et al., 2009; North et al., 2011; North & Pfefferbaum, 2013). One of the most important contributions of Tajima et al. (2024) is its examination of disaster-related experiences and problem drinking through the analysis of longitudinal data. With few exceptions (e.g., Pietrzak et al., 2012), much of what is known about disaster and behavioral and psychiatric health outcomes relies on postdisaster cross-sectional data, often limited to single-timepoint assessment. By using a longitudinal approach, Tajima et al. (2024) report a temporally sustained increase in problem drinking as assessed by the CAGE questionnaire (Bush et al., 1987) over the years of follow-up. This study represents an important step in understanding the dynamic processes underlying the effects of disaster experiences on individuals' problem drinking behaviors over time. Despite the study's strengths of using a longitudinal design, one notable limitation is the lack of data on participants' predisaster alcohol use and disorder status. This absence precluded the identification of, for example, onset of problem drinking or postdisaster relapse. Furthermore, this limitation is especially critical in view of the evidence that only a small number of individuals develop a new alcohol use disorder postdisaster (North et al., 2011). Thus, this emphasizes the necessity of conducting studies with longer intervals. More generally, while disasters offer a unique, natural opportunity to study how individuals respond to stressors, analyses relying only on postdisaster data are not meant to be interpreted causally. One approach to strengthen inferences in observational epidemiological data is to integrate comprehensive pre- and postdisaster information. This allows for delineation of temporal trajectories and patterns of drinking in response to various disaster-related stressors and experiences. Such an approach enables the characterization of individuals who develop new alcohol use disorder, those who have "resolved" from clinically significant problem drinking, individuals with persistent or recurrent alcohol problems, and those exhibiting complex lapsing and relapsing patterns. However, obtaining predisaster data is inherently challenging due to the unpredictable nature of disasters themselves. These events often occur unexpectedly, making it difficult to develop study protocols quickly while balancing the need to minimize burden among those affected by the disasters. To address this barrier, researchers could explore alternative data sources or methodologies. For example, asking participants to retrospectively report on their alcohol use behaviors could offer insights into their predisaster alcohol use patterns. Additionally, harnessing electronic health records would provide a means to infer predisaster information and psychopathology based on available data, without relying on additional extensive interviews or detailed questionnaires. Another strategy to address the lack of predisaster data challenge is to identify and collaborate with studies already in progress before a disaster and pivot to include new measures to address research topics of interest. In addition to the lack of predisaster data, Tajima et al. (2024) focused exclusively on CAGE-assessed problem drinking with no consideration given to levels of alcohol consumption. Broadening the assessment to encompass both consumption measures and symptomology of alcohol use disorders is imperative (North et al., 2013) for a comprehensive understanding of the impact of disaster on alcohol use dynamics over time. Gaining insight into the temporal dynamics and patterns, as well as predictors and correlates associated with various patterns is critical for informing the development and implementation of effective intervention strategies and supporting the diverse pathways to reducing harmful alcohol use and recovery from alcohol use disorder (Tucker et al., 2020). Another highlight of the Tajima et al. study (2024) is the pattern of results demonstrating the temporal dynamics and different impacts of various disaster-related experience or events on problem drinking over time. For instance, while encounters with life-threatening danger and experiences of discrimination or slurs were both associated with problem drinking, the effects differed in terms of magnitude and duration. Individuals who faced life-threatening danger showed a sustained increase in the likelihood of problem drinking throughout the 3-year study period, underscoring the enduring effect of such traumatic experience. Conversely, the effect of discrimination or slurs on problem drinking decreased over time, indicating a potential attenuation of these negative effects. Furthermore, Tajima et al. (2024) found that individuals who suffered major property loss or experienced the death of a colleague did not initially show differences in problem drinking compared with their counterparts at baseline; however, they exhibited an increase in likelihood of problem drinking over the subsequent 2 years. These results appear to provide some empirical support for theoretical perspectives positing different patterns of disruption in individuals' functioning across time (e.g., chronic disruption; delayed disruption) following potentially traumatic events and related experiences (Bonanno, 2004; Norris et al., 2009). Additionally, these findings are consistent with a robust literature linking the severity of disaster exposure and related adversities with varying levels of risk for behavioral health problems and psychopathology (Goldmann & Galea, 2014). Importantly, while the findings of Tajima et al. (2024) are intriguing, a key limitation of this study is that it estimated a series of univariable associations between disaster-related experience and problem drinking. In other words, each model examined a single disaster-related experience or stressor. This analytic approach precluded the evaluation of the unique and specific associations between different disaster-related experiences and problem drinking. This gap is significant because disasters are widely recognized as complex events, encompassing many interrelated disaster-related experiences or stressors (Bonanno et al., 2010). Conducting separate models may overlook the nuanced interplay among different stressors and aspects of disaster experiences. An important next step is to evaluate a dose–response effect of increasing exposure to disaster-related experiences on problem drinking. Modeling dose–response is of great interest because it directly addresses critical questions such as whether there is a linear (or nonlinear) increase in risk with increasing exposure to stressors and related experiences, and how much excess risk exists at different exposure levels. Understanding the dose–response effect has important scientific, intervention, and policy implications. Equally important is considering the interrelation among the disaster-related experience (unique effects). Analytically modeling the complex interplay among these experiences allows us to understand how stressors and disaster-related experiences interact and accumulate over time. Examining the unique effects of different stressors and disaster-related experiences could inform the development of tailored interventions that address specific needs. For example, interventions designed to mitigate the effects of discrimination experiences may differ from those targeting loss. Beyond merely documenting the association between disaster-related experience and course of problem drinking, several additional avenues for future research and next steps are essential for advancing our understanding of the interplay and temporal dynamics underlying common disaster-related experiences, stressors, and alcohol use behaviors among individuals affected by disaster. First, understanding the mechanisms through which disaster-related experiences influence the risk of harmful alcohol use is important. Future studies could explore factors such as coping strategies, social support and isolation, psychological resilience, and other mental health factors (Galea et al., 2005), as potential mechanisms. A mechanistic understanding of the associations between disasters and alcohol use behaviors will establish a foundation for designing more tailored intervention programs that target specific risk pathways. Second, there is considerable heterogeneity in individual responses to disaster exposure and related experiences and their associated impacts (Bonanno et al., 2010; Bryant, 2019). Notably, not all individuals exposed to disaster and related stressors develop clinically significant alcohol problems. Extensive research highlights various risk factors for postdisaster behavioral and mental health outcomes, including sociodemographic background and preexisting vulnerabilities such as mental health conditions, family history of psychiatric disorders, prior trauma exposure, and adversities (Bromet et al., 2017; Cerdá et al., 2011). This wide array of risk factors underscores the importance of identifying plausible contributors to variability in individuals' responses to disasters. For example, in the study by Tajima et al. (2024), similar pattern of results was observed between individuals with and without preexisting psychopathology. However, the lack of difference between the two groups may be attributed to the low prevalence (1%) of preexisting psychopathology within the sample. Furthermore, the study did not explore the potential moderating role of predisaster alcohol use/misuse. However, extant research indicates that increased postdisaster problem drinking often represents an exacerbation of preexisting alcohol misuse rather than the onset of new alcohol use disorders (Kleber & Van Der Velden, 2009; North et al., 2013). Thus, an important next step involves testing the hypotheses that the associations between disaster-related experiences and postdisaster problem drinking are more pronounced among individuals with preexisting alcohol problems or a family history of alcohol problems. Investing potential moderators can aid in identifying individuals at heightened risk for problem drinking following disasters. This knowledge is crucial for designing interventions aimed at mitigating risks among vulnerable populations. Third, there is a critical need for more research that focuses on understanding individuals' broader contextual factors, such as macro, sociocultural, and family/interpersonal relationship contexts. Most disaster studies primarily examine person-level variables, and it is surprising that the wider environmental context is often overlooked, given that individuals are embedded within multiple contexts. Understanding how individuals cope with disasters and their subsequent behavioral health outcomes requires considerations of this broader context (Bonanno et al., 2010; Hoffman & Kruczek, 2011). For instance, at a more macro, community level, the study by Tajima et al. (2024) indicated differences in the associations between specific disaster-related experiences and problem drinking among workers from the two nuclear plants (Daiichi and Daini). On a more micro level, research shows that support from romantic partners and features of marital and couple relationship provide important context for postdisaster functioning and recovery (Reid & Reczek, 2011), and social support contributes to a decrease in alcohol use (Cerdá et al., 2011). Thus, contexts of individual functioning and recovery require increased consideration within disaster—behavioral health studies, including alcohol outcomes. In summary, longitudinal findings, such as those reported in Tajima et al. (2024), enhance our understanding of how common disaster-related experiences, stressors, and alcohol use behaviors unfold over time among individuals affected by disasters. Utilizing a longitudinal approach and incorporating both pre- and postdisaster measures of alcohol use behaviors can advance our understanding of the temporal dynamics on how disaster exposure, related stressors, and experiences influence drinking patterns over time. Further research on mechanisms and moderators will improve our understanding of individuals most vulnerable to harmful alcohol use and alcohol-related problems following a disaster. Ultimately, this knowledge may inform the development of interventions and support programs aimed at reducing problem drinking and promoting recovery from alcohol-related problems in the aftermath of disasters. This work was supported in part by the National Institutes of Health Grants R01 AA028064 (PI: Salvatore), R01 AA030996 (PI: Salvatore), and R01 AA030563 (PI: McCutcheon). None.