Approximately 100,000 women annually seek emergency care after sexual assault in the United States alone [1]. More than half of these women develop persistent posttraumatic stress disorder (PTSD) symptoms; however, the vast majority do not receive any additional health care services outside of the emergency care visit [2]. PTSD and commonly comorbid symptoms, including anxiety, depression, pain, and substance use, impose a large burden of suffering on sexual assault survivors [3]. In light of both the personal suffering experienced by sexual assault survivors and the substantial societal costs (estimated at US $2.8 trillion over the lifetimes of women survivors [4]), the development of scalable secondary preventive interventions to reduce PTSD development after sexual assault is a public health priority.

One important cognitive process implicated in PTSD development after sexual assault is anxiety sensitivity (AS). Individuals with elevated AS, or fear of anxious arousal [5], may interpret normative posttraumatic reactions, such as intrusive memories, physiological reactions to trauma reminders, and hypervigilance, as threatening and something to be avoided [6,7]. Consistent with this theoretical notion, AS has been empirically demonstrated to be elevated among those with PTSD and to prospectively predict PTSD onset [8,9], course [10], and severity [11], including among sexual assault survivors presenting for emergency care [12].

Importantly for preventative intervention development, AS can also be modified by intervention. Brief computerized cognitive behavioral interventions have been shown to reduce AS [13,14], and, in turn, PTSD [15-17] and related symptoms such as anxiety and depression [13,14]. These interventions provide psychoeducation, including the adaptive nature of anxiety, and cognitive behavioral strategies, such as cognitive restructuring and interoceptive exposure. Such interventions have been demonstrated to be acceptable to clinical samples of community individuals with elevated AS [18].

Recently, we developed and conducted initial acceptability and feasibility testing of an AS intervention (RISE [RCT for Innovating Stress-Related E-Health] Guide) to reduce the development of PTSD in adult sexual assault survivors [19]. The RISE Guide uses methods with demonstrated use in previous AS interventions, including psychoeducation, cognitive restructuring, and interoceptive exposure. However, the content has been modified to be smartphone-based and specifically tailored to target PTSD symptoms (eg, psychoeducation is focused on the stress reaction after trauma and its functional use). Furthermore, the audio-visual and interactive content of the RISE Guide includes an ecological momentary assessment or intervention [20] component, which provides as-needed reminders of intervention concepts. This novel AS intervention was tested for proof of concept and acceptability among 12 women seeking emergency care after sexual assault [19]. The results indicated that 75% of participants initiated the intervention, with the majority (66.6%) completing most of it [19]. Participants rated the RISE Guide as logical and useful, perceiving it to be potentially helpful in reducing their symptoms [19]. In addition, the results showed moderate-to-large reductions in both AS and PTSD symptoms over 6 weeks post assault [19]. Therefore, AS interventions in general, and this specific, newly developed intervention, show promise as potential scalable interventions to be provided to women seeking emergency care after sexual assault.

The ORBIT model of behavioral intervention development offers a systematic framework to guide the development and testing of promising behavioral interventions to prevent chronic diseases [21]. Specifically, following the acceptability and proof-of-concept evaluations (phases 1 and 2a) [21], such as those described above [19], the next step in assessing promising interventions, such as the RISE Guide, is conducting a pilot study (phase 2b). A pilot study is designed to replicate the findings of clinically significant benefits from the active intervention and to assess the adequacy of the chosen control group prior to moving on to a pilot feasibility trial (phase 3). Of note, this study was initially designed as a “preliminary efficacy trial,” but given increasing recognition of the problems with these types of designs [22-25], it was shifted to a phase 2b pilot study. Therefore, this study was designed to address these next steps in intervention development by replicating proof-of-concept testing of the novel AS intervention (“RISE Guide”) and evaluating (1) intervention acceptability, (2) behavioral risk factor (AS) and symptom (PTSD) change, (3) acceptability or credibility of the chosen control intervention (“Breathe2Relax” [B2R], a digital app teaching relaxation techniques such as diaphragmatic breathing) and its impact on AS [26], and (4) the feasibility of a larger randomized controlled trial (RCT) testing the novel AS intervention among women presenting for emergency care after sexual assault (ie, whether 3‐4 participants per month could be recruited, >70% of the sample could be retained, and >70% would complete the RISE Guide).

The current phase 2b pilot randomized study of 60 women sexual assault survivors presenting to emergency care tested (1) the replication of acceptability, (2) the proof of concept of a novel smartphone-based intervention targeting AS (RISE Guide) in reducing AS and PTSD symptoms, (3) the adequacy of a relaxation control condition (B2R), and (4) feasibility of conducting future efficacy-focused RCTs. As expected, given the high-risk nature of sexual assault and the elevated AS in the development of PTSD symptoms [8,12,37,38], this was a vulnerable sample with 88.8% (40/45) meeting the clinical cutoff of 33 on the PCL-5 for likely PTSD 7 weeks post assault and 80.6% (29/36) continuing to meet this cutoff at month 6 [30]. These rates are consistent with previous research, including large observational cohort studies [2,39]. Overall, individuals receiving the RISE Guide rated it as acceptable and were observed to have medium-to-large reductions in AS and PTSD by month 6, particularly for those who completed the intervention. However, similar effects were found for those randomized to B2R, and we were unable to track how much participants engaged in B2R, limiting our ability to draw conclusions. Recruitment and retention were found to be feasible; however, there were concerns regarding intervention completion for those randomized to the RISE Guide. There were also major concerns about the appropriateness of the control condition, given that it produced larger effects on AS and PTSD symptoms than expected.

With regard to replicating the acceptability and proof of concept of the RISE Guide, as expected, the results indicated the RISE Guide was rated as acceptable via a validated measure (the CEQ) [35] and a study-specific measure. This is consistent with prior research indicating the acceptability of digital AS interventions generally [18], as well as the RISE Guide intervention specifically [19]. Replicating prior research [19], the results indicated that the RISE Guide was associated with medium-to-large reductions in AS and PTSD symptoms by month 6. However, in comparison to our initial trial, participants in this study experienced smaller and nonsignificant reductions in symptoms at week 7. By month 6, the mean PCL-5 score was approximately 44 (clinically significant according to lower cutoffs of 33, but not higher cutoffs of 45) [30,32]. Reductions were also clinically significant (including those in the control condition). We examined per-protocol analyses, given that intervention completion was lower than expected, and found that reductions in AS and PTSD symptoms were relatively larger in the group that completed the intervention versus those who did not. Yet, it must be acknowledged that (1) individuals who complete behavioral interventions may be more motivated and thus more likely to change regardless of intervention content and (2) the natural recovery of PTSD symptoms is to be expected for many sexual assault survivors [37]. Therefore, these analyses cannot pinpoint whether the intervention itself caused symptom reductions or whether they were related to individual traits or natural recovery. Furthermore, given that significant effects were not found immediately post intervention, we cannot determine whether the intervention took longer than expected to impact outcomes, or if by month 6, natural recovery had occurred and thus accounted for the significant reductions found. This confound highlights the need for future fully powered RCTs that can rule out recovery effects as an explanatory factor for symptom reductions.

Regarding feasibility, the results suggested that recruitment was feasible, with 3 participants enrolled per month. This was in line with study goals. Retention was also feasible, with at least 60% (37/62) completing all follow-ups, including month 6. This is notable considering that retention rates within samples of women sexual assault survivors presenting for emergency care have been particularly challenging, with rates often ranging from about 50% to 70% [40,41]. EMA adherence rates were about 50%, suggesting the EMA protocol may have been burdensome for participants to complete. Furthermore, there was a large range of compliance, with some participants completing very few and some demonstrating nearly perfect compliance. To place this adherence rate in context, relatively lower EMA adherence rates have been found in prior research among sexual assault survivors [42]. However, about 50% adherence is somewhat disappointing and should be improved in future trials.

As hypothesized and necessary for adequate control conditions [43], participants also rated the control condition (B2R) as acceptable. In contrast to the hypothesis, however, and problematic for an efficacy trial, B2R participants also experienced medium-to-large reductions in AS. Although effect sizes were relatively smaller than in the RISE Guide condition, they were still medium-to-large and larger than desired for a control condition. B2R was selected as a control condition, as we hypothesized that it would result in reductions in PTSD, but to a lesser extent than the active condition [44]. We did not hypothesize that it would significantly change AS, but the current findings suggest that it did. However, it must be noted again that we cannot rule out the effects of natural recovery on reductions in AS. Even with this in mind, the current results suggest B2R may not be an appropriate control condition. In particular, it is difficult to estimate how many women in our sample would be expected to develop PTSD without any treatment, as there are no estimates to our knowledge available for how many women sexual assault survivors with high AS develop PTSD. The current sample demonstrated high rates of PTSD across both conditions, with 80.6% (29/36) continuing to meet criteria for probable PTSD at month 6. This is similar to an unselected sexual assault survivor sample [2,39] in an observational study, making it unclear whether our intervention was ineffective or if this group of women with high AS would have had even higher rates of PTSD without intervention. While the severity of PTSD symptoms highlights the importance of preventative interventions in this group, it also makes it difficult to assess intervention success without a benchmark of recovery rates in this high-risk group. Furthermore, there is currently no standard intervention provided at the time of emergency care to sexual assault survivors to prevent PTSD [45]. Thus, the inclusion of an active control is not aligned with the actual clinical experience of sexual assault survivors seeking emergency care, who do not systematically receive any active intervention for PTSD.

Given these issues, it may be appropriate to consider “treatment-as-usual” as a control condition in future studies. Current guidelines for SANE examinations require trauma-informed care, informed consent, a comprehensive physical assessment, evidence collection, medical care (eg, prophylactic treatment for sexually transmitted infections and pregnancy), and documentation. Currently, unlike physical needs, there is no standardized system of mental health risk assessment and prevention as part of the examination; however, SANEs must provide trauma-informed care, ensure physical or psychological safety, and provide referrals to appropriate follow-up care [46]. Using treatment-as-usual as a control would reflect the clinical reality on the ground, allow the collection of estimates of natural recovery among women sexual assault survivors with high AS, and better enable testing of whether our novel intervention holds promise for its intended use in emergency care. Indeed, experts have raised concerns about the selection of overly stringent and unrealistic active control conditions, particularly in the early phases of research, as these can (potentially inaccurately) reduce enthusiasm about promising novel interventions [43]. Treatment-as-usual would also appropriately reflect a relatively less formidable control that is appropriate for early-stage RCTs, such as this work.

In terms of active intervention feasibility, one-third of the participants did not initiate any component of the RISE Guide intervention. On the other hand, 59% (16/27) completed the entire intervention. This was consistent with our initial trial, in which 25% did not initiate the intervention [19]. It is expected that not all participants will complete behavioral interventions [47], given the motivation, time, and access to logistics required. Indeed, traditional interventions, such as cognitive processing therapy or prolonged exposure, have widely varying completion rates, which are often found to be around 40% to 50% [48-50]. Completion rates are particularly poor for digital health interventions, which tend to have low levels of engagement [51]. Given this context, the current completion rate of nearly 60% (16/27) is promising and within the range of gold-standard interventions. However, 33% (9/27) of the participants not initiating the intervention is concerning in the context of an RCT and may lead to problems with hypothesis testing in an efficacy trial. One way to overcome problems with intervention adherence is to ensure we are targeting the appropriate audience for our intervention [47]. A common approach to doing so is the addition of a “behavioral run-in” period mimicking what participants would be asked to do during the intervention, and only randomizing individuals who are able to successfully complete the behavioral run-in period [47]. Therefore, we suggest that future research consider using a behavioral run-in period similar to the intervention procedures, such as completing daily phone-based tasks, in order to ensure the appropriate target population for the intervention.

Overall, this study aligns with previous findings that cognitive behavioral secondary preventative interventions may be beneficial in mitigating PTSD after sexual assault. However, this is a challenging area in which proper RCT design is critical, given the natural recovery observed after trauma exposure. Specifically, future research should continue to test this potentially promising intervention by selecting a more appropriate control condition, such as treatment-as-usual, to better approximate the current clinical reality and allow for the establishment of typical rates of recovery in high-risk samples. Furthermore, using a behavioral run-in period to help ensure that the correct target sample is identified and treatment engagement is sufficiently high may be helpful. The results suggest that individuals with high AS are a vulnerable group after sexual assault, but it remains unclear whether targeting AS will lead to reductions in PTSD symptoms after sexual assault.

The results of the current study should be considered in light of its strengths and limitations. Strengths include that the RISE Guide was tested in the target population, and intervention completion was assessed objectively. Limitations include, first, as discussed, about a third of those randomized to the RISE Guide did not initiate the treatment. Second, we were also unable to assess how many of those randomized to B2R initiated or completed that intervention, so it is unclear how these rates compare. Therefore, we are unable to interpret any between-group comparisons or comment on the feasibility of using B2R as a control condition in future work. Third, although consistent with a phase 2b trial [21], the sample size was relatively homogeneous and limited. As research progresses, larger and more diverse samples (eg, men and gender diverse sexual assault survivors) should be included. Fourth, larger sample sizes would also enable direct statistical comparisons between the active and control groups’ clinical outcomes. Fifth, larger sample sizes would allow for the calculation of more reliable effect sizes. Effect sizes derived from pilot RCTs such as this one should be interpreted with caution [24].

This pilot study of 60 women sexual assault survivors tested the feasibility and replicated the proof of concept of a novel smartphone-based intervention targeting AS versus a relaxation control to reduce AS and, in turn, PTSD symptoms. The findings reinforced the vulnerability of this population, with high rates of PTSD observed 6 months after the sexual assault. Acceptability and proof-of-concept findings replicated prior research [19] and suggest that more research is needed, particularly given the larger than expected effects of the control condition and the need to rule out recovery effects. This work is essential to the development of effective scalable early interventions for this common trauma population that experiences great suffering.