The day-to-day relationship between posttraumatic stress symptoms and social support after sexual assault

ABSTRACT

Background: Experiencing sexual assault is associated with a significant increase in risk for developing posttraumatic stress disorder and related concerns (e.g. alcohol misuse). Cross-sectional and longitudinal evidence suggests that social support may be both broadly protective against and eroded by posttraumatic stress symptoms. However, little is known about how different aspects of social support and posttraumatic stress symptoms influence each other in the weeks and months immediately following sexual assault, when posttraumatic stress (PTS) symptoms first emerge.

Objective: The present study assessed the day-to-day relationship between social support and PTS in a sample of distressed, alcohol-using, recently-assaulted female survivors participating in a clinical trial of an app-based intervention (N = 41).

Method: Participants completed 3 weeks of daily diaries starting within 10 weeks of sexual assault. Mixed-effects models were used to examine prior-day and same-day relationships between PTS and four social support constructs (social contact, emotional support, pleasantness of social interactions, and talking about sexual assault).

Results: Results indicate that higher quantity and pleasantness of social interactions over the full sampling period was associated with lower PTS symptoms on any given day. Experiencing better-than-typical social interactions on one day was associated with lower than typical PTS symptoms on that day and the next day. On days when participants discussed their sexual assault with others, they tended to be having higher than usual PTS symptoms.

Conclusions: Findings suggest that increasing the quantity and pleasantness of social interactions soon after sexual assault might protect against worsening posttraumatic stress symptoms.

Trial registration: ClinicalTrials.gov identifier: NCT03703258.

HIGHLIGHTS

• In N = 41 distressed and alcohol-using female survivors of recent sexual assault, having a higher quantity of social interactions and more pleasant social interactions within 10 weeks of assault was associated with lower posttraumatic stress symptoms.

• When survivors’ social interactions were more pleasant than average on one day, their posttraumatic stress symptoms tended to be lower than average the next day, and recent survivors were more likely to talk about sexual assault on days when their posttraumatic stress symptoms were higher than usual.

• Interventionists should take note that increasing the quantity and pleasantness of social interactions soon after sexual assault might protect against worsening posttraumatic stress symptoms.

Antecedentes: La experiencia de agresión sexual se asocia con un aumento significativo en el riesgo de desarrollar un trastorno de estrés postraumático y preocupaciones relacionadas (por ejemplo, abuso de alcohol). Evidencia transversal y longitudinal sugiere que el apoyo social puede ser tanto un amplio protector, como también verse deteriorado por los síntomas de estrés postraumático. Sin embargo, poco se sabe sobre cómo diferentes aspectos del apoyo social y los síntomas de estrés postraumático se influyen mutuamente en las semanas y meses inmediatamente posteriores a una agresión sexual, cuando los síntomas de estrés postraumático (SEPT) aparecen por primera vez.

Objetivo: El presente estudio evaluó la relación diaria entre el apoyo social y los SEPT en una muestra de mujeres sobrevivientes angustiadas, usuarias de alcohol, recientemente agredidas, que participaron en un ensayo clínico de una intervención basada en una aplicación (N = 41).

Método: Las participantes completaron 3 semanas de registros diarios, comenzando dentro de las 10 semanas posteriores a la agresión sexual. Se utilizaron modelos de efectos mixtos para examinar las relaciones del día anterior y del mismo día entre los SEPT y cuatro constructos de apoyo social (contacto social, apoyo emocional, nivel de agrado en las interacciones sociales y hablar sobre la agresión sexual).

Resultados: Los resultados indican que una mayor cantidad y agrado de las interacciones sociales durante todo el período de muestreo, se asociaron con menores SEPT en un día determinado. Experimentar interacciones sociales mejores de lo habitual, en un día, se asoció con SEPT inferiores a lo habitual en ese día y al día siguiente. En los días en que las participantes conversaron de su agresión sexual con otros, tendían a tener SEPT más elevados de lo habitual.

Conclusiones: Los hallazgos sugieren que aumentar la cantidad y el nivel de agrado de las interacciones sociales poco después de una agresión sexual, podría proteger contra el empeoramiento de los síntomas de estrés postraumático.

KEYWORDS:

• Sexual assault
• posttraumatic stress
• social support
• daily diary
• acute post-trauma period

PALABRAS CLAVE:

• Agresión sexual
• estrés postraumático
• apoyo social
• registros diarios
• período agudo post-trauma

Sexual assault (SA), defined as unwanted and nonconsensual sexual contact, is pervasive and costly to societies and individuals (Risser et al., 2006; Smith et al., 2018; Stotzer, 2009). Survivors are at heightened risk for a range of mental health consequences, including posttraumatic stress (PTS; Dworkin, 2020; Kelley et al., 2009; Kessler et al., 1995; Kessler et al., 2014; Kilpatrick et al., 2003). In the month following SA, most survivors exhibit PTS symptoms (Dworkin et al., 2021), and many also exhibit comorbid alcohol misuse (Dworkin, 2020). Because most survivors experience remission of these early symptoms within the first months following the assault (Gutner et al., 2006; Rothbaum et al., 1992), posttraumatic stress disorder is often conceptualized as a ‘failure to recover’ from a typical stress response (Dworkin et al., 2021). Identifying factors contributing to recovery soon after assault for distressed substance-using survivors may inform the development of interventions to facilitate these recovery processes.

Interpersonal processes, and social support in particular, may affect recovery processes. Social support is one of the strongest correlates of posttraumatic stress (PTS) in cross-sectional research (Brewin et al., 2000; Ozer et al., 2003): Survivors with higher PTS tend to have less social support (Vogt et al., 2017; Wagner et al., 2016).

However, the direction of the relationship between social support and PTS – i.e. whether PTS erodes social support, social support protects against PTS, or both – remains unclear. The ‘social erosion’ model purports that PTS causes disturbances in interpersonal relationships which eventually leads to a decline in social support (Carter et al., 2016; Hall et al., 2014; Kaniasty & Norris, 1993; Price et al., 2014; Wagner et al., 2016). In contrast, ‘social causation’ model suggests that strong social support protects against PTS (Hobfoll et al., 1990; Johansen et al., 2022; Littleton et al., 2022; Thoits, 1986).

Evidence for these models has been mixed. Some longitudinal studies have found support for the social erosion model but not the social causation model (Carter et al., 2016; Hall et al., 2014; King et al., 2006; Laffaye et al., 2008). Other longitudinal studies have found support for the social causation model but not the social erosion model (Freedman et al., 2015). Still others have found evidence for a bidirectional relationship between social support and PTS: A recent meta-analysis of longitudinal studies found that social support and PTS reciprocally predict each other, with similar effect sizes in both directions (Wang et al., 2021).

These mixed findings may be due in part to methodological issues. First, studies have varied widely in time between trauma exposure and data collection, yet little attention has been paid to time since trauma exposure as a potential determinant of whether the social erosion versus social causation model is supported. Social causation as it was originally conceptualized – i.e. with social support protecting against worsening PTS symptoms – would be expected to occur in the weeks and months immediately following trauma (Wagner et al., 2016), whereas social erosion would presumably occur later once social resources are chronically strained or depleted. Indeed, of the four studies reviewed that enrolled participants within 16 weeks of a potentially traumatic event, three offer support for the social causation model (Freedman et al., 2015; Johansen et al., 2022; Perry et al., 1992) and only one offers support for the social erosion model (Price et al., 2014). Given there is theoretical and empirical reason to believe that time since trauma exposure may influence the extent to which social erosion or social causation occurs, additional attention should be paid to this variable to clarify the relationship between social support and PTS.

Second, most investigations on social support and PTS have focused on how the two constructs relate to each other over the course of months or years. However, if PTS does, in fact, erode social support over time and vice versa, we might expect to see associations between these two variables on a day-to-day basis. The only daily diary study on this topic found support for the social causation model, and also identified a relationship not observed at larger timescales – namely, a positive association between PTS and social support over time in a sample of college-aged women with past-month PTSD and lifetime SA assessed daily for 28 days (Dworkin et al., 2018). Accordingly, additional daily diary studies of survivors with elevated symptoms are needed in the early post assault period to better understand the developmental processes by which PTS and social support might influence each other.

Lastly, ‘social support’ has been conceptualized in different ways across studies. Studies differ in the focus of support: some studies focus on trauma-specific support (e.g. social reactions to disclosure), while others focus on general support (e.g. the number of friends one has). Studies also differ in the measurement of support: some studies have examined the quantity of social contact (i.e. how often trauma survivors talk to their friends, e.g. Hall et al., 2014), while others have examined the types of support behaviors received in these interactions, including emotional (e.g. Dworkin et al., 2018; Platt et al., 2016), informational (e.g. King et al., 2006; Price et al., 2014), or tangible support (e.g. Perry et al., 2021); still others have assessed survivors’ perceptions of support availability or support interactions, regardless of the actual behaviors present (Johansen et al., 2022). Few studies have directly compared these patterns across multiple types of support, and in those that have, meaningful differences have emerged by support type (e.g. Platt et al., 2016). Since so few studies have compared multiple types of support, there is no discernable pattern of association between support operationalization and findings (erosion versus causation) across studies.

1.1. The current study

Little is known about the developmental processes by which PTS and social support influence each other in the early aftermath of SA. Examining these questions among distressed, substance-using survivors could help to clarify these dynamics in a particularly vulnerable population. To address this gap, we conducted a secondary analysis of daily diary data on four aspects of social support and PTS in distressed, alcohol-using, female survivors of recent SA participating in a clinical trial of an app-based intervention. We examined a range of social support constructs that reflected differences in both the focus of the support (SA-specific versus general/global) and the way that the support construct was measured (interaction quantity versus support behaviors received versus perceptions of interactions): (1) quantity of social interactions (quantity of general/global support), (2) quantity of interactions in which SA was discussed (quantity of SA-specific support), (3) emotional support received (received general/global support), and (4) perceived pleasantness of social interactions (perceived general/global support). Our goal was to examine evidence for the social erosion and causation models in the early post-trauma period in an exploratory manner and clarify which specific social support constructs were uniquely associated with PTS.

2. Method

The present study was a secondary analysis from a pilot clinical trial of an mHealth early intervention for recent SA survivors with elevated PTS and alcohol use (N = 41). Participants in completed daily diaries assessing PTS and social support over a 21-day period starting within 10 weeks of SA. The trial was registered on www.ClinicalTrials.gov (NCT: NCT03703258) and procedures received approval from the University of Washington Institutional Review Board (see: Dworkin et al., in press).

2.1. Procedures

2.1.1. Eligibility criteria

Participants were eligible if they were female, experienced nonconsensual and unwanted sexual contact within the last 10 weeks, were ≥18 years old, were fluent in English, had smartphone and internet access, had ≥1 alcoholic drink in the past month and ≥1 episode of high-risk drinking in the past 6 months (>3 drinks in one day or >7 drinks in one week), and endorsed elevated symptoms of PTS. ‘Elevated symptoms of PTS’ was defined as sufficient items as ‘moderately’ or above within at least 3 of 4 symptom clusters (≥1 intrusions item, ≥1 avoidance symptom, ≥2 negative alterations in cognition and mood items, and/or ≥2 hyperarousal items) on the PTSD Checklist for DSM-5, a 20-item self-report measure assessing PTSD symptom severity (PCL-5; Weathers et al., 2013). Provisional PTSD diagnostic status was not examined as a component of study eligibility given the recency of the traumatic event. However, we descriptively characterized the number of participants meeting provisional diagnostic criteria, defined as endorsing sufficient items as ‘moderately’ or above in all 4 symptom clusters (The National Center for PTSD (n.d.)). Suicidality with intent or psychosis were exclusion criteria.

2.1.2. Recruitment

Participants were recruited via emails sent to the student body at the University of Washington, community flyering in public spaces (e.g. gyms, cafes, salons) in the greater Seattle area, social media advertisements (targeted for 18+ age range and female gender), and referrals from therapists and other community-based care providers in the greater Seattle area over the course of seven months, from January through August 2021.

2.1.3. Screening and consent

Interested participants (n = 174) contacted the study team, 74% of whom (n = 128) were assessed for eligibility via phone screen. Of those who completed a phone screen, 39% (n = 50) were eligible to complete the baseline, 38% (n = 49) completed the baseline, and 32% (n = 41) met symptom criteria for randomization. N = 41 were then automatically randomized and granted access to the app.

2.1.4. Intervention

The THRIVE intervention is a coached mHealth smartphone application for survivors of recent SA. It includes daily cognitive–behavioral exercises (e.g. cognitive restructuring, activity scheduling), optional relational skills modules, a daily survey assessing PTSD symptoms and alcohol use behavior, and weekly calls with a coach over 21 days. To isolate the effects of the cognitive–behavioral exercises and relational skills modules, participants were randomized to receive either all components of the THRIVE intervention or a control version, which included the daily surveys and weekly coaching calls only, and no cognitive–behavioral exercises or relational skills modules. Relational skills modules were infrequently used by participants. See Dworkin et al. (2023) for additional details.

2.1.5. Daily surveys

Participants completed a brief daily survey in the THRIVE app for 21 days. Participants reported on the period from 6:00 am on the prior day to 6:00 am on the current day. Participants indicated (Yes/No) whether they had texted/DMed, talked on the phone, spent time in person, or had a video call with ‘one or more friends, family members, romantic partners, or other people you have social relationships with’; any ‘yes’ response indicated the presence of social contact. If participants endorsed social contact, they were asked about the perceived pleasantness of social interactions (‘Overall, how were these interactions?’; 1 = very unpleasant, 7 = very pleasant), received emotional support (‘During any of these interactions, did you receive emotional support (such as listening to you, comforting you) related to any issue you were experiencing?’; 0 = No, 1 = Yes), and whether they talked about the assault during their social interactions (‘During any of these interactions, was your unwanted sexual experience or your reactions to it discussed?’; 0 = No, 1 = Yes). Daily PTS symptom severity was assessed using the 4-item version of the PCL-5 (Price et al., 2016), modified for daily use. These items assessed ‘Repeated, disturbing, and unwanted memories of the unwanted sexual experience,’ ‘Avoiding external reminders of the unwanted sexual experience,’ ‘Having strong negative beliefs about yourself, other people, or the world,’ and ‘Feeling jumpy or easily startled.’

2.2. Participants

Participants were N = 41 adults (mean age = 20.78, SD = 3.40) assigned female sex at birth. Participants were White (43.9%), Asian/Asian American (14.6%), Black/African American (12.2%), or multiracial (29.3%); 14.6% were Latinx. About half were non-heterosexual (34.1% bisexual, 7.4% gay or lesbian, and 2.4% other). Most (90.3%) were students. Participants’ SA occurred an average of 3.50 weeks (SD = 2.23 weeks; range = 0, 9) prior to enrollment. At baseline, 82.3% of participants met criteria for provisional PTSD (absent the 1-month symptom duration criterion) on the PCL-5 using the cluster-specific criteria described in the Eligibility criteria section.

Participants completed an average of 15.85/21 days of self-monitoring (SD = 4.97; range: 2–21 days; total days across participants: 650). Survey completion was correlated with average PTS symptom severity across monitoring days, r = −0.40, p < .01, and average pleasantness of social interactions, r = 0.32, p = .045, such that participants with lower PTS symptoms and more pleasant social interactions across monitoring days tended to complete more surveys (rs = −0.40 and 0.32, respectively). Because our analyses examined lagged relationships between prior-day and same-day variables, we retained only the 501 days of self-monitoring on which a survey had also been completed the prior day. There were no significant differences in days with or without a prior-day survey as a function of focal study variables.

2.3. Analyses

All analyses were conducted in R (R Core Team, 2021). We analyzed prior-day and same-day scores on measures of PTS and social interactions. For all primary analyses, we tested mixed-effects models with random intercepts because scores were nested within participants. Same-day scores were used as dependent variables. We created two types of independent variables to separate between- and within-person variance: (1) the mean of each participant’s scores across self-monitoring days (i.e. person mean) and (2) the difference in each day’s score from the person mean (i.e. daily deviation). For binary variables, the person mean represents the proportion of days that that variable was present, and the daily deviation represents whether the variable was present on a given day.

For each social support variable, we conducted separate models to predict (1) same-day PTS from same- and prior-day social support deviation, prior-day PTS deviation, and person-mean social support and (2) same-day social support from same- and prior-day PTS deviation, prior-day social support deviation, and person-mean PTS. Study condition and time since SA (person mean and daily deviation) were included as covariates. Person-mean time since SA reflected the time since SA at the midpoint of each participant’s self-monitoring period, the time since SA daily deviation variable reflected whether participants were earlier or later in their daily self-monitoring period on a given day. For clarity, we refer to the time since SA daily deviation variable as ‘study day.’

Models predicting binary variables (i.e. presence/absence of social contact, emotional support, and talking about the assault) were analyzed using mixed-effects logistic regression via the glmer function of lme4 package (Bates et al., 2015). For models predicting count variables (i.e. PTS symptom severity, perceived pleasantness of social interactions), we explored the need to use generalized linear mixed models with Poisson or negative binomial distributions using likelihood ratio tests. We determined that a linear mixed model (using the lmer function of lme4) was indicated for the model predicting perceived pleasantness of social interactions and a generalized linear mixed model with a negative binomial distribution (using the glmer.nb function of lme4) was indicated for the models predicting PTS.

2.4. Sensitivity analyses

We (1) examined differences in focal variables by condition and (2) re-ran each primary model for each condition separately.

3. Results

All participants had social contact on at least one day, and social contact occurred on M = 93.4% of participants’ monitoring days (SD = 11.8%, range: 47.6% to 100%). Of the days that participants had social contact, they received emotional support on M = 37.7% of days (SD = 25.6%, range: 0.0% to 95.0%), rated the pleasantness of their interactions as M = 5.62/7 (between ‘a little pleasant’ and ‘somewhat pleasant’) (SD = 0.83, range: 4.15–6.95), and talked about the SA on M = 11.9% of days (SD = 18.3%, range: 0.0% to 87.5%). Model results predicting social support variables are presented in Table 1, and results for models predicting PTS are presented in Table 2.

Table 1. Models predicting social interaction variables.

Independent variables	Study model (dependent variable: social support variables)
	Presence of social contact model (n = 501) B (SE)	Presence of received emotional support model (n = 496) B (SE)	Perceived pleasantness of social interactions model (n = 437) B (SE)	Talking about sexual assault model (n = 497) B (SE)
Intercept	5.50 (1.25) p = 1.23e−05	−1.26 (0.63) p = .046	1.85 (0.06) p < 2e−16	−4.46 (1.00) p = 8.72e−06
Same-day PTS symptom deviation	0.21 (0.17)	0.07 (0.12)	−0.04 (0.02)	0.70 (0.20) p = .001
Prior-day social interaction variable	0.18 (0.11)	−0.06 (0.11)	−0.01 (0.02)	−0.07 (0.14)
Prior-day PTS symptom deviation	−0.18 (0.17)	−0.07 (0.12)	−0.01 (0.02)	0.00 (0.20)
Person-mean PTS symptom severity	−0.28 (0.14) p = .046	0.01 (0.09)	−0.01 (0.01)	0.12 (0.12)
Time since sexual assault	−0.01 (0.19)	−0.44 (0.26)	−0.01 (0.01)	−0.48 (0.37)
Study day	−0.60 (0.22) p = .009	0.06 (0.13)	0.01 (0.02)	0.01 (0.20)
Condition	0.42 (0.87)	0.80 (0.54)	−0.03 (0.04)	1.35 (0.75)

Table 2. Models predicting PTS symptom severity.

Independent variable	Study model (dependent variable: PTS)
	Presence of social contact model (n = 501) B (SE)	Presence of received emotional support model (n = 496) B (SE)	Perceived pleasantness of social interactions model (n = 437) B (SE)	Talking about sexual assault model (n = 497) B (SE)
Intercept	2.85 (0.60) p = 9.61e−07	1.55 (0.15) p < 2e−16	2.79 (0.55) p = 2.44e−07	1.53 (0.11) p < 2e−16
Prior-day PTS symptom deviation	0.06 (0.02) p = .005	0.05 (0.02) p = .011	0.06 (0.03) p = .016	0.06 (0.02) p = .006
Same-day social interaction deviation	0.02 (0.02)	0.01 (0.03)	−0.09 (0.02) p = 5.17e−05	0.09 (0.02) p = 1.57e−05
Prior-day social interaction deviation	−0.03 (0.02)	0.03 (0.02)	−0.05 (0.03) p = .013	−0.02 (0.02)
Person-mean social interaction	−1.46 (0.63) p =0.024	−0.01 (0.32)	−0.22 (0.09) p = .020	0.13 (0.48)
Time since sexual assault	0.04 (0.07)	0.04 (0.04)	0.03 (0.07)	0.04 (0.08)
Study day	−0.10 (0.03) p = 4.40e−05	−0.11 (0.03) p = 9.41e−05	−0.10 (0.03) p = .000	−0.10 (0.03) p = 4.38e−05
Condition	0.19 (0.15)	0.11 (0.16)	0.03 (0.15)	0.08 (0.16)

3.1. Presence of social contact

First, we tested bidirectional relationships between PTS and the presence of social contact in the full sample of self-monitoring days (N = 501).

In the model predicting presence of social contact, person-mean PTS was negatively associated with presence of social contact: Participants with higher person-mean PTS were less likely to have social contact on any given day (β = −0.28; Table 1; Figure 1a). Study day was significantly negatively associated with the presence of social contact, such that participants were less likely to have social contact as their self-monitoring period progressed (β = −0.60; Table 1; Figure 1(a)).

Figure 1. (a–d) Model results divided by social support variable.

In the model predicting PTS, person-mean presence of social contact was significantly negatively associated with PTS symptoms: Participants who had more frequent social contact across the self-monitoring period tended to have lower PTS on any given day (β = −1.46; Table 2; Figure 1(a)). Study day was significantly negatively associated with PTS, indicating that PTS tended to decrease as the self-monitoring period progressed (β = −0.10; Table 2; Figure 1(a)). Prior-day PTS symptom deviation was significantly positively associated with same-day PTS, meaning that people tended to have higher PTS when their prior-day PTS had been higher than typical (β = 0.06; Table 2; Figure 1(a)).

3.2. Presence of received emotional support

Second, we tested bidirectional relationships between PTS and receiving emotional support in a sample of n = 496 (5 days were excluded from the full sample of 501 via listwise deletion due to missing data). Emotional support was coded as absent on days with no social contact. We conducted sensitivity analyses in the subsample of days with social contact only, and results were consistent.

In the model predicting presence of emotional support, no significant relationships were identified (Figure 1(b)). In the model predicting PTS, study day was significantly negatively associated with PTS (β = −0.11; Table 2; Figure 1(b)), and prior-day PTS symptom deviation was significantly positively associated with PTS (β = 0.05; Table 2; Figure 1(b)).

3.3. Perceived pleasantness of social interactions

Third, we tested bidirectional relationships between PTS and pleasantness of social interactions. These analyses were conducted in the subset of days with social contact (n = 461); 25 days were excluded via listwise deletion due to missing data on the social interaction variable, for a final sample of n = 436 days.

In the model predicting pleasantness of social interactions, no predictors were significantly associated with pleasantness of social interactions (Figure 1(c)). In the model predicting PTS, person-mean pleasantness of social interactions, prior-day deviation in pleasantness of social interactions, and same-day deviation in pleasantness of social interactions were negatively associated with PTS (β = −0.22, β = −0.05, and β = −0.09 respectively; Table 2; Figure 1(c)). This suggests that people who rated social interactions as more pleasant on average tended to have lower PTS on any given day, and higher-than-typical ratings of the pleasantness of social interactions on one day were associated with lower PTS on that same day and the next day. As with the other models predicting PTS, study day was significantly negatively associated with PTS (β = −0.10; Table 2; Figure 1(c)), and prior-day PTS symptom deviation was significantly positively associated with PTS (β = 0.06; Table 2; Figure 1(c)).

3.4. Talking about sexual assault

Finally, we tested bidirectional relationships between PTS and whether participants reported talking about their SA. These analyses were tested in the full sample of self-monitoring days (N = 501) with 4 days excluded via listwise deletion due to missing data on the social interaction variable. Talking about SA was coded as absent on days without social contact. We conducted sensitivity analyses in the subsample of days with social contact only, and results were consistent.

In the model predicting talking about SA, daily PTS symptom deviation was significantly positively associated with talking about SA: People with higher-than-typical PTS on a given day were more likely to report having talked about their SA with others on that day (β = 0.70; Table 1; Figure 1(d)).

In the model predicting PTS, talking about SA on a given day was significantly positively associated with higher PTS: People who talked about SA on a given day were more likely to report having higher PTS on that day (β = 0.09; Table 2; Figure 1(d)). As with all other models predicting PTS, study day was significantly negatively associated with PTS (β = −0.10; Table 2; Figure 1(d)), and prior-day PTS symptom deviation was significantly positively associated with PTS (β = 0.06; Table 2; Figure 1(d)).

3.5. Comparison by condition

We tested bivariate differences by condition in person-mean values of focal variables. Intervention-condition participants talked about SA more (t = −3.053, p = .002) and had more social interactions (t = −2.65, p = .008) compared to participants in control. Control-condition participants had social interactions that were significantly more pleasant than intervention-condition participants (t = 2.62, p = .009). There were no significant differences in PTS or emotional support between conditions.

We then tested each of the primary models for each condition separately (Supplementary Tables 1 and 2). Although we did not find any substantive differences in the relationship between social support and PTS, two differences in model results by condition emerged. First, for control (but not intervention) participants, mean time since SA was negatively associated with talking about SA (Supplementary Table 1). In other words, control-condition participants for whom more time had elapsed since SA were less likely to discuss SA on any given day. The same effect was not found in the intervention-only model or the model with all participants. Second, for intervention (but not control) participants, social interaction on one day was associated with an increased likelihood of social interaction on the following day (Supplementary Table 1).

4. Discussion

This study examined day-to-day relationships between PTS and four aspects of social support starting within 10 weeks following SA: (1) the presence of social interaction, (2) perceived pleasantness of social interactions, (3) emotional support received during social interactions, and (4) whether or not participants discussed SA in their social interactions. Our findings offer the first evidence for the social causation model on a daily scale and in the months immediately following SA, when symptoms are first emerging and thereafter rapidly changing for many survivors (Dworkin et al., 2021).

4.1. Prospective relationships: social causation vs. erosion?

A main goal of this study was to examine differences in relationships between PTS and social support as a function of the type of social support examined. This is important because social erosion versus causation might operate differently for different types of social support. Our findings suggest that such differences may indeed be present. When examining prospective relationships (i.e. between variables from one day to the next), which can potentially speak to social erosion versus causation, we found only that higher social interaction pleasantness (but not quantity of social interactions, receipt of emotional support, or talking about sexual assault) was associated with lower next-day PTS symptoms, but PTS was not prospectively associated with any social support variables. This provides evidence for social causation (but not erosion) with regard to social interaction pleasantness only, and it suggests that the perceived pleasantness of a social interaction might protect against worsening PTS in the acute post-SA period. These findings are in line with past work suggesting that social causation is likely to occur soon after trauma exposure and that erosion is less likely to occur during this same period (Freedman et al., 2015; Johansen et al., 2022; Perry et al., 1992; Price et al., 2014).

Interestingly, prior prospective studies starting in the early post-trauma period showed evidence for social causation with regard to other forms of social support, including received support (defined as receiving guidance, emotional support, and/or tangible support) (Littleton et al., 2022) and perceived support (defined as attachment, social integration, nurturance, reassurance of worth, reliable alliance, and guidance) (Johansen et al., 2022). However, neither of these studies examined the presence of these relationships on a day-to-day basis, which might account for the differences seen in this study.

The lack of a significant prospective relationship between PTS and emotional support differs from the only prior study day-to-day relationships between emotional support and PTS, which found that higher PTS on one day predicted higher-than-average received emotional support the following day in female college-aged SA survivors with past-month PTS (Dworkin et al., 2018). This finding was notable because it did not reflect either the social causation or erosion model. However, this relationship was not observed in the current study. One potential explanation for this discrepancy in findings is that data in the present study were collected starting within 10 weeks of SA, whereas the prior study did not exclusively enroll recent survivors. It is possible that a positive relationship between PTS on one day and emotional support on the following day is not present in the months immediately following assault, but instead, develops over time.

4.2. Concurrent relationships

All other model relationships found were concurrent, and thus cannot speak to social erosion versus causation. However, they still illuminate differences in relationships to PTS across types of social support.

4.2.1. Same-day relationships

Regarding same-day relationships between variables, lower social interaction pleasantness and talking about the assault were associated with higher same-day PTS symptoms, but neither presence of social interaction nor presence of emotional support was associated with same-day PTS severity. One potential explanation is that lower social interaction pleasantness and talking about the assault both might either affect or be affected by transient negative mood states associated with higher-than-typical PTS, whereas a survivor’s friends might reach out (i.e. initiate a social interaction) and offer emotional support regardless of a survivor’s PTS on a given day.

4.2.2. Relationships involving average levels of social support or PTS

Regarding relationships involving average levels of social support, we found that people who had more days of social interaction or perceived their interactions as more pleasant on average tended to have lower PTS on any given day, but neither number of days with emotional support nor number of days when the assault was discussed were associated with PTS severity on any given day. This suggests that survivors with lower PTS tend to maintain regular social engagement and a positive general outlook on their social interactions, whereas survivors might regularly receive emotional support or frequently talk about the SA regardless of their PTS symptom severity. These findings confirm a general relationship between PTS symptoms and both the quantity of and perceptions of social interactions, but it does not clarify whether social erosion versus causation is operating for these types of social support during this time period. Indeed, one possible explanation for this is the social causation theory, i.e. that having more social support or more pleasant interactions alleviated PTS (Flint et al., 2020). Alternatively, it is possible that survivors with lower PTS may have been less likely to avoid social interaction or that survivors with higher PTS may have found that others were less willing to engage in social interactions with them (i.e. social erosion).

4.3. Condition-based differences

In testing bivariate differences by condition in average values of focal variables, we found that intervention-condition participants talked about SA more, had more social interactions, and perceived their interactions as more pleasant compared to participants in the control group. These differences are likely explained by participants in the intervention group having received daily prompts to engage in positive and valued behaviors and reframe negative cognitions. Although they also had access to optional relationship-skills modules, these modules were infrequently used by participants (Dworkin et al., 2023). However, in sensitivity analyses testing each of the primary models for each condition separately, we found no differences in the relationships between social support variables and PTS variables (Supplementary Tables 1 and 2). This finding suggests that the conclusions drawn from our primary models – models which include data from all participants and control for condition – are defensible.

Of note, we did find two between-condition discrepancies, both unrelated to PTS. First, we found that for control (but not intervention) participants, there was an association between time elapsed since SA and likelihood of discussing SA: Those for whom more time had passed since SA were less likely to discuss SA on any given day (Supplementary Table 1). It is possible that the intervention prevented a decrease in discussions of the SA. Second, for intervention (but not control) participants, we found that social interaction on one day was associated with an increased likelihood of social interaction on the following day (Supplementary Table 1). The source of this discrepancy is unclear but could have been due to the cognitive–behavioral intervention modules completed by those in the intervention condition. Specifically, intervention condition participants may have used activity scheduling intervention modules to plan a series of social interactions, which could have created this auto-regressive effect of social interactions for the intervention condition participants. While these between-condition discrepancies are important to note and discuss, they do not suggest the presence of remarkably different relationships between PTSD and social support variables between conditions and therefore do not put into question the primary results.

4.4. Strengths & limitations

The present study had several strengths. First, we examined relationships between social support and PTS on a day-to-day basis – a smaller time scale than has been used in most prior work on this topic – thereby offering new insights into how these constructs relate to each other over time. Indeed, the day-to-day effect of PTS on social support and vice versa is likely quite relevant to clinical interventions. For example, our finding that higher perceived pleasantness of social interactions on one day is associated with lower PTS symptoms the following day indicates that interventions that bolster survivors’ ability to have pleasant social interactions soon after assault may facilitate survivors’ recovery. Second, this study was the first to examine social support-PTS dynamics soon after SA, thereby advancing our understanding of how social support contributes to recovery processes (Gutner et al., 2006; Rothbaum et al., 1992). The present study is also unique in that it examined several types of social support in a single study, which allowed us to draw conclusions regarding the differential relationships of various types of social support and PTS. Finally, findings from the present study are generalizable to survivors with elevated PTS and substance use and could therefore be leveraged to inform clinical intervention development relevant to this especially vulnerable survivor population.

Findings should be interpreted in the context of several limitations. First, participants were recruited for having elevated PTS symptoms and alcohol use, all participants were female, and most were college students; accordingly, findings may not be representative of survivors’ experiences more broadly. Second, although sensitivity analyses indicated no substantive differences in findings between control and intervention conditions, intervention activities may have changed the behaviors and experiences of those in the intervention condition, thereby influencing the results. Third, the components of social interactions assessed were measured with unvalidated questions, two of which were dichotomous and therefore could have obscured meaningful variation in experiences of social contact. Fourth, participants could complete surveys about the prior day at any point in a 24-hour window, which may have affected recall accuracy. Fifth, participants may have been less likely to respond to surveys on days when their PTS were highest – indeed, participants with lower PTSD symptoms and more pleasant social interactions across the sampling period tended to complete more surveys. Accordingly, the true range of social support and PTSD symptom severity may not be adequately reflected in the data. Finally, our sample was relatively small and had moderate missing data.

Future work on this topic should investigate the day-to-day relationships between the four social support constructs examined in the present study and PTS symptoms in a larger and more demographically diverse sample of SA survivors to investigate the extent to which the present findings are generalizable. Certainly, components of social support not measured herein are also important to SA survivors’ long-term mental health trajectories and should be studied. For example, there are additional types of received and perceived social support and of SA-specific social interactions not measured in the present study (e.g. received and perceived tangible support, received SA-specific emotional support, perceived pleasantness of SA-specific social interactions). Such components may be then leveraged by clinical interventionists to expedite recovery. Similarly, times scales both smaller (hour-to-hour) and larger (week-to-week) should be considered in future work to empirically establish the ‘speed’ of the relationship between social interactions and PTS symptoms and inform the timing of related interventions.

5. Conclusion

In sum, results suggest that having more social interactions and perceiving them to be more pleasant on average may be associated with lower PTS on any given day and vice versa in recent SA survivors. Findings provide support for the social causation model on a day-to-day basis in the months immediately after assault, and they suggest that survivors may be more likely to talk about SA on days when their PTSD symptoms are higher than usual.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data are not publicly available at this time.

Additional information

Funding

Research reported in this publication was supported by the National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under [grant number R00AA026317]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.