An estimated one in five women are sexually assaulted while attending college, exposing them to heightened risks of adverse mental health outcomes (Conley et al., 2017; Muehlenhard et al., 2017). Prospective studies have found that 41.9% of sexual assault survivors meet PTSD criteria one  year after the assault (Dworkin et al., 2023). A recent meta-analysis further concludes that individuals with a history of sexual assault have over seven times greater odds of developing PTSD compared to those with no such history (OR = 7.57; Dworkin, 2020). Child sexual abuse also increases risk of PTSD and increases risk of future victimization among college women (Messman-Moore et al., 2000). Sexual victimization and PTSD have been associated with increased risk of alcohol misuse (i.e., alcohol cravings and consumption) among college women, where each predicts the other over time (Read et al., 2012).

Survivors of sexual victimization (SV)—defined in this paper to encompass both childhood sexual abuse and sexual assault during adolescence and adulthood—also face challenges in regulating both positive and negative emotions (Berfield et al., 2022). Emotion dysregulation in PTSD can be seen via symptoms of persistent negative affect and difficulty experiencing positive affect (Hofmann et al., 2012; Berfield et al., 2022). Additionally, the college years are often a period of elevated and dynamic stressors that may cause fluctuations in mood (Zhang and Zheng, 2017; Barker et al., 2018).

College years and young adulthood are marked by high rates of alcohol consumption (Merrill and Carey, 2016; Krieger et al., 2018). Generally, college students tend to drink more often and more heavily than non-college-attending peers (Merrill and Carey, 2016). College drinking can be motivated by multiple factors including drinking to cope with negative affect, drinking to fit in, drinking to improve a good time or celebrate, and drinking as part of a social connection (Baer, 2002; Park and Levenson, 2002; Osberg and Boyer, 2018; King and Valley, 2019). As such, although individuals with sexual victimization histories, and those with PTSD, may drink to reduce symptoms or manage negative affect, they may also drink for other reasons as well (Huh et al., 2015; O’Donnell et al., 2019). College students also drink to cope with factors aside from PTSD including academic stress, negative mood, and affect variability (Gottfredson and Hussong, 2013; Peterson et al., 2021).

The prevailing explanatory theory behind relationships between SV, PTSD, and alcohol misuse is the self-medication theory, where alcohol is consumed as a means to cope with aversive symptoms of PTSD and to manage affect-related disturbances (Khantzian, 2003; Rhew et al., 2017; Dardis et al., 2021; Gilmore et al., 2024). There is ample cross-sectional, longitudinal, and micro-longitudinal or daily-level research supporting relationships between elevated PTSD symptoms and increased alcohol use and problems (Smith and Cottler, 2018; Hawn et al., 2020). Research that has focused on same-day or next-day relationships between PTSD symptoms and alcohol use has generally found that increased PTSD symptom severity or increased variance in PTSD symptom severity predicts same-day drinking (Cohn et al., 2014; Kaysen et al., 2014; Hruska et al., 2017; Wilson et al., 2017).

There are limitations to this general body of research. Many of the studies conducted to date use a PTSD total symptom severity score, which does demonstrate general positive associations between overall PTSD symptom severity and alcohol use (Hruska and Delahanty, 2012; Delker and Freyd, 2014; Tuliao et al., 2016; Eddinger et al., 2019). However, this may mask contributions from specific symptoms in relation to drinking. For example, in studies that have examined symptom clusters or individual symptoms some have found that re-experiencing symptoms or avoidance/numbing symptoms (Jakupcak et al., 2010; Lee et al., 2015) predict alcohol use but have not found those same relationships with arousal symptoms (Maguen et al., 2009; Jakupcak et al., 2010; Simpson et al., 2012; Hellmuth et al., 2013; Kaysen et al., 2014; Lee et al., 2015; Langdon et al., 2016). Among daily studies there have been several that have looked in more nuanced ways at how clusters or individual symptoms of PTSD relate to same-day or next-day drinking. Simpson et al. (2012) found that intrusive and avoidance symptoms were associated with next-day alcohol cravings in a small community sample. Kaysen et al. (2014) showed that female college students with elevated intrusive and behavioral avoidance symptoms experienced stronger urges to drink and were more likely to consume alcohol on those days. Sullivan et al. (2020) found that each PTSD symptom cluster was uniquely associated with concurrent alcohol consumption in women who had experienced intimate partner violence.

Affect also plays a critical role in understanding drinking behavior. Affect can be thought of as occurring along two dimensions, that of valence or pleasantness or unpleasantness and arousal or the level of activation and energy (Russell and Barrett, 1999; Posner et al., 2005). Two meta-analyses of non-clinical samples have found positive affect to be associated with near-term alcohol use, but no such association was found for negative affect (Tovmasyan et al., 2022; Dora et al., 2023). Among community members, higher positive affect was associated with increased alcohol consumption but only increased arousal and decreased variability in arousal predicted the likelihood of drinking (Peacock et al., 2015). Mean level of arousal predicted the quantity of alcohol consumed. In another community study, days with higher-than-usual levels of high arousal positive affect were associated with an increased odds of alcohol consumption and of heavy episodic drinking whereas there were no significant associations between high or low arousal negative affect and alcohol use (Jones et al., 2021). Conversely, research on women exposed to violence has shown that negative affect predicts same-day urges to drink and that affect regulation plays a crucial role in this link (Kaysen et al., 2007, 2014; Dyar and Kaysen, 2023; Stappenbeck et al., 2023).

These findings highlight the temporal and complex relationships between PTSD symptoms, affect, and drinking behavior, lending empirical support to the self-medication hypothesis. However, there remains a significant gap in the literature regarding the short-term longitudinal stability of PTSD symptoms, affect, and drinking behavior, such as day-to-day fluctuations over the course of a week or month. Research shows that PTSD symptoms are not constantly present among those with the diagnosis or subthreshold symptoms but instead can vary across time (McFarlane, 2000; Solomon et al., 2009) and even fluctuate widely from 1 day to the next (Black et al., 2016; Biggs et al., 2019; Schuler et al., 2021). The interactions among PTSD symptoms can be conceptualized as a short-term dynamic system to determine how variables affect one another over brief time intervals (Bringmann et al., 2023). To our knowledge, only one study to date has used a short-term dynamic model to examine PTSD symptoms (Greene et al., 2018), and no studies have yet assessed longitudinal changes in affect and alcohol misuse, nor the short-term dynamics between these measures. Thus, existing literature focuses on whether overall PTSD symptom severity or negative affect predict alcohol misuse, but does not test the role of fluctuations in PTSD severity or affect in understanding drinking. Understanding the role of these fluctuations could inform more precise interventions and help identify periods of increased vulnerability to alcohol misuse. This is the difference between whether people with higher PTSD or higher negative affect are at higher risk of alcohol misuse versus whether individuals who have more variability in their affect and symptoms have higher risk of alcohol misuse.

An additional limitation of the extant literature on relationships between PTSD and affect is the question regarding the time span within which these factors may influence drinking behavior. Existing micro-longitudinal studies examining links between PTSD-related symptoms and drinking behaviors have focused on PTSD or affect predicting drinking within the same assessment (so essentially micro-cross-sectional) or predicting the next assessment. Although there are theoretical justifications for examining same-day and next-day associations, it is equally important to investigate the extent to which changes in PTSD symptoms or affect may result in longer-term changes in drinking behaviors, as these may be mechanisms important in understanding development of alcohol use disorders over time. Without research addressing lagged effects, including longer term lagged effects, we cannot accurately validate our theoretical models concerning the temporal dynamics of how PTSD symptoms influence drinking behaviors.

The Institutional Review Board of the University of Washington approved all procedures, and we collected written consent from all participants.

We conducted all statistical analyses in SPSS Statics 28. We initially conducted a Kruskal–Wallis test to assess potential variations in completions rates for daily assessments among the three groups. Results did not reveal significant differences in completion rates among the three groups (p = 0.295). Completion rates were comparable, with the PTSD group at 68.2 ± 0.25%, the SV-exposed group at 69.3 ± 0.27%, and the no-trauma group at 75.0 ± 0.24%. We proceeded with the main data analysis.

The overall goal of this study was to explore the impact of SV on short-term, longitudinal stability and immediate and time-lagged predictive dynamics of PTSD symptoms, affect, and drinking behaviors among college women. To address this goal, we first compared the longitudinal stability of these measures across three groups: college women with a history of SV and PTSD, those with a history of SV but no PTSD, and those with no history of either. We then applied support vector regression to determine which feature set of PTSD symptom networks and affect best predicts drinking behavior at the optimal specific time lag within the 0–7 day range.

Prior to addressing goals 1 and 2, we conducted validation analyses to confirm anticipated data trends regarding PTSD symptoms, affect, and drinking behavior among the groups. As expected, our findings indicated significant group differences, with the PTSD group showing the highest mean levels of PTSD symptoms, arousal, negative affect, and alcohol cravings, and the lowest levels of positive affect compared to the SV-exposed and no-trauma groups. These findings confirm previous research that has highlighted the emotional and psychological impact of PTSD on individuals (Peter-Hagene and Ullman, 2015; Straus et al., 2018). As is consistent with extant literature, PTSD is associated with higher arousal, negative affect, and alcohol cravings, as compared to SV exposure alone, and compared to college women without prior trauma exposure. SV also has an effect, but less than SV coupled with PTSD.

The first goal of this study was, in part, to examine the longitudinal stability of PTSD symptom networks, comparing across those with PTSD, those with sexual victimization but without PTSD, and those without a trauma history. Looking at the stability of these networks across these groups allows us to test whether these symptoms are more chronic or more transient based on SV history and probable diagnosis. Results indicate greatest longitudinal stability of PTSD symptom networks among participants with probable PTSD, with lowest variance across the metrics for women in the PTSD group and highest variance for the no-trauma controls. Consistent with prior research showing stability of PTSD symptom networks over longer periods, ranging from six months to five years (Von Stockert et al., 2018; Ge et al., 2019; Crowe et al., 2023), our findings are the first to show micro-longitudinal stability of PTSD symptoms using a network analysis approach. In contrast, using linear mixed models, Biggs et al. (2019) compared daily fluctuations in PTSD symptom severity across two groups of U.S. military personnel – those with and without probable PTSD. Results showed that among personnel with probable PTSD, symptoms were more severe on weekdays relative to weekends. No such variance was observed among those without probable PTSD suggesting more variance for those with PTSD as compared to more stability as was found in our study. These contradictory findings may be attributable to differences in methodology (i.e., analytic approach and/or operationalization of the disorder), college women as compared to military personnel, type of trauma exposure, or the presence of binge drinking. For example, the authors note that the weekday context of working in a military environment may expose service personnel to frequent reminders of military-related traumas. Conversely, college women with sexual assault histories may encounter less day-dependent, reminders of their traumatic experiences. Given the limited amount of research in this area, and the differences in findings across these two studies, replication is warranted. Our results suggest that the temporal stability of symptoms among individuals with PTSD can be attributed to the persistent and enduring nature of the disorder, whereas individuals without trauma histories exhibit greater fluctuation in symptom correlations, indicating that these responses may be indicative of momentary distress to life stressors.

Additionally, the PTSD group exhibited greater longitudinal stability in arousal and greater longitudinal variance in negative affect compared to both comparison groups, indicating significant challenges in emotional regulation. This pattern aligns with existing literature that links PTSD to emotional dysregulation (Weiss et al., 2020; Haws et al., 2022). The combination of stable, persistent arousal symptoms and fluctuating negative affect highlights a complex emotional profile, consistent with previous research associating the arousal cluster with emotional dysregulation (Weiss et al., 2020; Haws et al., 2022). Notably, difficulty in maintaining persistent arousal and regulating mood (negative affect) is associated with maladaptive coping strategies such as alcohol use (Lee et al., 2015; Von Stockert et al., 2018; Ge et al., 2019; Haws et al., 2022; Crowe et al., 2023).

With respect to drinking behavior, our findings showed that the PTSD group had higher longitudinal variance in alcohol cravings and consumption compared to the other groups. These findings suggest that individuals with PTSD experience greater longitudinal variability in alcohol cravings and consumption. This variability suggest that individuals with PTSD may crave and use alcohol as a coping mechanism to manage their persistent PTSD symptoms and arousal, as well as fluctuating mood states (Bolton et al., 2009; Turner et al., 2018). Our combined findings of longitudinal stability of PTSD symptoms and arousal, higher longitudinal variance of negative affect (mood), and higher longitudinal variance of alcohol cravings and use indicate a complex interaction between stable distress and fluctuating mood states in individuals with PTSD. This pattern underscores the chronic and multifaceted nature of PTSD, where persistent symptoms are coupled with attempts to self-medicate through alcohol use to manage emotional dysregulation. These insights highlight the need for targeted interventions that address both the enduring symptoms of PTSD and the maladaptive coping strategies employed by individuals to alleviate their emotional distress.

Regarding our second goal, we found that alcohol cravings and consumption were best predicted by unique PTSD symptom network features (i.e., correlated sets of PTSD symptoms) experienced 3–4 days prior. Specifically, intrusive memories, concentration problems, distress at reminders, physiological arousal at reminders, emotional numbing, and sleep disturbance optimally predicted alcohol cravings 3 days later. Intrusive memories, irritability, intrusive memories, hypervigilance, psychogenic amnesia, and loss of interest best predicted alcohol consumption 4 days later. Notably, four of the six factors predicting alcohol consumption or cravings included at least one symptom from the re-experiencing cluster. This is consistent with existing research demonstrating associations between re-experiencing symptoms and drinking outcomes (Simpson et al., 2012; Kaysen et al., 2014; Sullivan et al., 2020). Unlike these previous studies, which found immediate (i.e., same or next assessment) effects, our results highlight the significance of time-lagged associations, particularly with arousal-cluster symptoms. The finding that certain PTSD symptom interactions may have a delayed effect on drinking behavior has important implications for treatment and relapse prevention strategies. For example, “urge surfing” is such a strategy used to manage cravings. The underlying model conceptualizes alcohol cravings and usage to cycle over short periods, typically within minutes to hours (Ostafin and Marlatt, 2008; Hisler et al., 2022). The strategy therefore encourages its users to ride out the wave of discomfort associated with an unmet urge to drink, suggesting that it will be short-lived. However, our findings suggest that some PTSD symptoms may have lingering effects on alcohol cravings and use. Similarly, in relapse prevention, clinicians work with clients to identify immediate triggers for potential use, but these findings suggest that the triggers may be more temporally distal. Ongoing research is needed to confirm and further explore dynamics of time-lagged effects within the self-medication hypothesis. For example, future research should test whether the stress of symptoms accumulates over a period of time, then resulting in drinking behaviors. Regardless of causal mechanisms, if confirmed, present findings may suggest that clinicians should work with clients regarding managing not only day-to-day stressors as antecedents of drinking but also addressing how to manage lingering effects of those stressors over time for those with PTSD.

Finally, while the PTSD group experienced higher alcohol cravings, there were no significant differences in actual alcohol consumption across the groups. This suggests that although individuals with PTSD have stronger urges to drink, these cravings do not necessarily result in increased consumption. Higher cravings may reflect an ongoing desire to use alcohol as a coping mechanism, consistent with the self-medication hypothesis, but actual drinking behavior may be influenced by situational constraints and individual differences in managing these urges.

Our choice of cohort, which included only individuals who already engage in heavy episodic drinking, may also contribute to the observed lack of differences in drinking consumption, as it did limit the sample to students already engaging in some high-risk drinking. However, this specific cohort provides unique insights into how trauma and PTSD symptoms specifically influence cravings within a context where high levels of drinking are more prevalent. Our results suggest that cravings may reflect underlying psychological distress and motivations for drinking that differ from actual consumption levels. Future research should further assess these dynamics to understand the underlying mechanisms and to develop targeted interventions.

Our findings on the longitudinal stability and time-lagged effects of PTSD symptom networks and effect on alcohol cravings and consumption have significant implications for interventions targeting college women exposed to SV. Understanding the specific PTSD symptoms and affective states that predict drinking behavior can lead to more targeted and effective therapeutic approaches. Cognitive-behavioral therapies that incorporate emotion regulation training and exposure therapy may help individuals develop healthier coping mechanisms and reduce their reliance on alcohol (Messman-Moore and Ward, 2014). Findings generally support the use of integrated trauma-focused substance use treatment, such as COPE as the skills can address reducing both the PTSD and managing alcohol cravings (Mills et al., 2012; Back et al., 2014). By focusing on the underlying emotional and psychological factors driving alcohol use, these interventions can provide more sustainable and effective outcomes.

Future studies should also develop computational methods to consider more than three features, which could reveal more complex interactions between symptoms and drinking behavior. While our current approach focuses on the three most important symptom correlations due to computational limitations, this strategy also ensures that therapy can be effective by prioritizing the most impactful factors. This dual focus on computational efficiency and clinical relevance highlights a promising direction for future research and therapeutic development.

The current investigation has several limitations that warrant consideration. First, analyzed data on PTSD and drinking was collected only once per day within a consistent two-hour window, preventing us from establishing the within-day temporal sequence of PTSD symptoms, affect, and drinking behavior. A strength of the present study is its goal of establishing the lag time in which correlations between PTSD symptom features and drinking outcomes were strongest rather than pre-supposing the temporal aspect of such associations as concurrent or immediately subsequent. The present method does, however, assume that symptoms experienced over a 24-h period affect subsequent 24-h periods, when within-day fluctuations or the time of day in which symptoms are experienced may have unique effects on drinking behavior. Fortunately, future research that overcomes this limitation is feasible. Prior micro-longitudinal studies have successfully collected data on PTSD and drinking behavior from four (Possemato et al., 2012) to eight (Gaher et al., 2014) times daily, and a review of this literature did not find that repeated micro-longitudinal monitoring, itself, increased or decreased alcohol use or PTSD symptoms (Lane et al., 2019). Scholarship on PTSD and alcohol use may greatly benefit from applying the present analytic approach to a larger sample with multiple, random same-day assessment points. The retrospective nature of recalling PTSD and drinking concurrently may have influenced participants’ responses. The study also did not assess daily drinking-related consequences. Although some negative consequences of drinking occur infrequently and are less likely to be a factor in micro-longitudinal studies, in cross-sectional and macro-longitudinal studies consequences are likely to show associations with negative affect and PTSD symptoms (Read et al., 2012; Tripp et al., 2015). Another limitation lies in the sample composition, which consists of women attending a competitive college and potentially excludes those with greater difficulty related to trauma history or drinking who might be unable to attend school. The study’s focus on women with a history of SV and a PTSD diagnosis who engage in higher risk drinking further limits generalizability to a broader population and limits the conclusions that can be drawn regarding the stability of PTSD symptoms to a higher drinking sample. These findings would need to be replicated within a broader sample of individuals with PTSD, regardless of drinking levels. The relatively small size of our sample limited our ability to account for demographic covariates and to investigate whether there were differential effects of childhood abuse versus sexual victimization during adolescence or adulthood (Walker et al., 2021; Boumpa et al., 2022). It is possible this sample was biased by the effects of PTSD among some women in our target population. PTSD symptoms such as avoidance of reminders and difficulty concentrating, may have been barriers to participation among women who were experiencing these symptoms more strongly. In addition, those with the most severe PTSD may drop out of college and never have the opportunity to be included in the research from college settings. Finally, data used in these secondary analyses were collected shortly before 2013, utilizing subjective mood assessments and measures of PTSD aligned with DSM-IV criteria. This limits the interpretation of findings in the context of the current DSM-5 criteria. Although there is generally good agreement between DSM-IV and DSM-5 findings (Rosellini et al., 2015; Crespo and Gómez, 2016), future research should aim to validate these findings using current assessments.

In conclusion, this study provides valuable insights into the short-term, longitudinal stability and predictive dynamics of PTSD symptoms, affect, and drinking behavior among college women exposed to SV. Our findings highlight the chronic and severe nature of PTSD symptoms and their significant impact on emotional regulation and alcohol cravings. The observed stability in PTSD symptoms and the variability in affect suggest the importance of targeting both persistent symptoms and emotional dysregulation in interventions. The identification of time-lagged associations between specific PTSD symptoms and drinking behavior, with delays of 3–4 days, contrasts with the immediate effects assumed in the self-medication hypothesis (i.e., same or next day). This underscores the complexity of these interactions and the need for comprehensive therapeutic approaches that consider longer-range effects. Additionally, our choice of cohort, which included individuals who already engage in binge drinking, suggests that cravings are a critical variable to assess in populations where drinking behavior is normalized due to social factors. Future research should expand on these exploratory findings by incorporating longer-range temporal assessments and exploring a broader range of symptom interactions. Addressing the psychological and environmental factors influencing alcohol use in individuals with PTSD can lead to more effective and tailored intervention strategies.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

The studies involving humans were approved by University of Washington Institutional Review Board. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

SB: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. MS: Conceptualization, Formal analysis, Writing – review & editing. TW: Conceptualization, Funding acquisition, Writing – original draft. MK: Resources, Supervision, Writing – review & editing. DK: Conceptualization, Data curation, Funding acquisition, Supervision, Writing – review & editing.