Psychotherapy research in the deaf community: pilot clinical trial lessons learned

Abstract

At the time of publication, there are no evidence-based psychotherapies to treat any behavioral health condition with Deaf clients. This article describes unique study design considerations for psychotherapy clinical trials conducted in the U.S. Deaf community. We synthesized emergent themes from participant exit interviews with feasibility data and real-life challenges that our team encountered when implementing the Signs of Safety pilot clinical trial, conducted from 2019 to 2022. Particularly illustrative participant accounts were selected to demonstrate five major lessons learned—expanding reach for recruitment; formally assessing participants’ ASL fluency; selecting ethically-sound control groups; streamlining video fidelity monitoring; and making crystallized outcome assessments ASL-accessible. These lessons learned informed the design of the first-ever full-scale psychotherapy trial in the U.S. Deaf community, to be conducted from late autumn 2024 through 2028. This trial will potentially validate the first evidence-based therapy for Deaf people and provide a vital roadmap for conducting Deaf community-engaged clinical trials.

Clinical or methodological significance of this article

The current article summarizes methodological lessons learned from a psychotherapy pilot clinical trial conducted in the New England Deaf community. The critical information we gained has since informed the design of the first-ever full-scale psychotherapy trial in the U.S. Deaf community—the Signs of Safety clinical trial—which will be conducted from late autumn 2024 through 2028. The Signs of Safety clinical trial will potentially validate the first evidence-based therapy for Deaf people and provide future behavioral health researchers with a vital roadmap for conducting community-engaged clinical trials with Deaf sign language users.

Background and objective

The objective of this article is to describe unique study design considerations for psychotherapy clinical trials conducted with Deaf sign language users. The U.S. Deaf community is a sociolinguistic minority group of approximately one million people who use American Sign Language (ASL) as a primary language (Mitchell & Young, 2022). Deaf sign language users experience systemic barriers to education (Cawthon et al., 2022; Lillo-Martin et al., 2023), employment (Dong et al., 2023), and health care access (Hall & Ballard, 2024; Jacob et al., 2022; Rotoli et al., 2022). These barriers underlie significant behavioral health disparities compared to the general population (Reiher, 2022). Documented disparities include, but are not limited to, triple the rate of lifetime problem drinking (Anderson et al., 2018; Berman et al., 2010; Rendon, 1992; Titus et al., 2008) and twice the rate of trauma exposure (Anderson et al., 2011; Anderson & Leigh, 2011; Black & Glickman, 2006; Porter & Williams, 2011; Schild & Dalenberg, 2012).

Hearing individuals have access to several validated treatments for comorbid trauma and alcohol use disorder. However, at the time of this publication, there are no evidence-based treatments to treat any behavioral health condition with Deaf clients (Anderson et al., 2021; Glickman & Pollard, 2013; NASMHPD, 2012). Available treatments fail to meet Deaf clients’ unique language access needs (Anderson et al., 2021; Glickman & Pollard, 2013). For example, Deaf people’s median English literacy level falls at the fourth grade (Morere, 2011) and health literacy gaps are common (Chandanabhumma et al., 2024; James et al., 2022b; McKee et al., 2019). Specifically, health-related vocabulary among Deaf sign language users parallels non-English-speaking U.S. immigrants (Barnett et al., 2011). Therefore, treatment resources designed for hearing populations require plain text revisions, filmed ASL translations, or education through storytelling to better match Deaf clients’ varied linguistic and cultural needs (Glickman, 2009; Kreuter et al., 2007; Slater & Rouner, 2006).

Leveraging extensive community engagement to address these barriers, we assembled a team of Deaf and hearing researchers, clinicians, filmmakers, actors, artists, and Deaf people in recovery to identify an optimal evidence-based therapy to adapt for Deaf clients and to collaborate on intervention development. The individuals comprising this team were primarily located in the state of Massachusetts, with most working for or consumers of Deaf mental health services at UMass Chan Medical School or the Massachusetts Department of Mental Health. All team members had previously recognized the lack of Deaf-accessible evidence-based treatments as a major barrier to Deaf mental health care and were excited to move this work forward.

The team was also staffed by one Deaf Community Advisor, a Deaf lay person with lived experience of PTSD and alcohol use disorder. In the past decade, our team has successfully worked with fourteen Deaf Community Advisors across four NIH-funded studies. Our study methods and dissemination products have been significantly improved and well-received by the community due to the invaluable guidance and contributions of our Deaf Community Advisors. These individuals have been recruited to the team using various strategies—social media campaigns, advertisements sent to collaborating agencies, and word of mouth. For the present study, our Deaf Community Advisor openly identified as a former participant of our center’s research on trauma and addiction and, following their study participation, expressed high levels of interest in continuing to support this research and joining the research team.

Together, this team developed Signs of Safety (Anderson et al., 2021). Signs of Safety is a culturally- and linguistically-accessible therapy toolkit for trauma and addiction that can be used with clients who are Deaf ASL users to supplement the Seeking Safety cognitive-behavioral counseling protocol (Najavits, 2002). Information regarding the development of the Signs of Safety toolkit and initial pilot testing (2014–2019) can be found in Anderson et al., 2021. For the purposes of this article, we present methodological considerations informed by feasibility data, qualitative participant feedback, and real-life challenges to study operation from a pilot feasibility clinical trial of Signs of Safety, conducted from 2019 to 2022.

Method

The Signs of Safety pilot feasibility study was a two-arm trial that randomized participants to immediately receive 12 weekly sessions of Seeking Safety + Signs of Safety therapy (experimental condition) or to receive this same intervention after a waiting period of 4 months (control condition). Participants were assessed at baseline, week 4, week 8, week 12/end of treatment, and week 16/1-month follow-up. Primary outcomes were past-month PTSD severity, as measured by the PTSD Checklist for DSM-5 (PCL-5), and past-month frequency and quantity of alcohol use, as measured by the Alcohol Timeline Followback.

It is important to note that the mid-2019 launch of this in-person psychotherapy trial was significantly disrupted by the onset of the COVID-19 pandemic in March 2020. In response to this unanticipated challenge, our team paused study operations, overhauled our in-person intervention methods, and implemented a virtual clinical trial—an acceleration of the inevitable development needed to scale Signs of Safety research to a national level. The methods described below reflect the virtual approaches our team implemented after the onset of the pandemic.

Recruitment and enrollment

We recruited participants using ASL YouTube videos and plain English flyers disseminated on Deaf listservs, Deaf community Facebook groups, and sent to local Deaf-specialized agencies. Recruitment materials were designed in collaboration with the team’s Deaf Community Advisor to increase engagement, perceived relevance, and comprehension among our study population. Recruitment materials directed interested individuals to contact the research team for eligibility screening and study enrollment.

Informed consent

Interested individuals were offered a 1-hour videophone appointment during which the principal investigator (co-author Anderson, a hearing ASL-fluent clinical psychologist) conducted informed consent procedures directly in ASL. Prior to the call, each participant was sent an electronic copy of the IRB-approved, written English informed consent form (e-Consent) via Research Electronic Data Capture (REDCap). During the call, Dr. Anderson presented each section of the e-Consent form in ASL (e.g., “What are the risks of being in this study?”, “What happens to information about me?”), pausing after each section for questions and discussion. Individuals who wished to enroll signed the e-Consent form in REDCap, by typing in their name or using the “wet signature” feature. Individuals who consented to participate were immediately screened by Dr. Anderson to determine study eligibility.

Inclusion and exclusion criteria

Individuals were eligible to participate if they self-identified as a Deaf ASL user and reported any amount of past-month alcohol consumption on the Alcohol Use Disorder Identification Test (AUDIT). Additionally, study enrollment required endorsement of past-month subthreshold or full PTSD on the PTSD Checklist for DSM-5 (PCL-5). Our operational definition of subthreshold PTSD was the endorsement of at least two of the four PTSD symptom clusters (re-experiencing, avoidance, negative cognitions and mood, arousal) at a severity of "moderate" or higher.

Individuals were excluded from the current study if they were younger than 18 years, unable to independently provide consent, were prisoners, or were pregnant. We did not formally assess pregnancy status; however, our team did not knowingly enroll anyone pregnant at the time of screening.

Participants were also excluded if they were engaged in concurrent formal individual psychotherapy—that is, if they would be receiving individual psychotherapy outside of the study simultaneous to receiving the Seeking Safety/Signs of Safety study intervention. Reasons for this exclusion criterion include the ethical dilemma of receiving individual psychotherapy from two different therapists simultaneously, as well as the research dilemma of not being able to determine if any observed treatment effect is due to the study intervention or the outside psychotherapy. Aligning with the Seeking Safety model, peer recovery meeting attendance (e.g., AA, NA) was encouraged and tracked as a potential mechanism of change.

Exclusion criteria were intentionally minimal to recruit a diverse sample. Given the high rates of comorbidity with both alcohol use disorder and PTSD, individuals with other behavioral health comorbidities (e.g., mood disorders, anxiety disorders, personality disorders, substance use disorders other than alcohol use disorder) were not excluded from our sample.

Study procedures

Randomization

Immediately following baseline assessment (described in Measures below), participants were randomized to one of two conditions: (1) a 12-session protocol of Seeking Safety supplemented with the Signs of Safety toolkit; or (2) an assessment-only waitlist control. Randomization was conducted within study therapist assignment and stratified by the participant’s self-reported gender. We assigned participants to one of four study therapists based on the participant’s stated preference for therapist gender, as well as their stated primary language/communication system (e.g., native ASL, Pidgin Signed English, Signed English, etc.).

Study conditions

Experimental condition

Experimental participants were offered 12 weekly 1-hour sessions of Seeking Safety with the Signs of Safety toolkit. Seeking Safety is an existing, present-focused, first-stage manualized treatment for trauma and addiction (Najavits, 2002). Session content engages clients in themes relevant to addiction and trauma, and to help them learn a specific skill to target symptoms of both addiction and PTSD (e.g., “Coping with Triggers,” “Asking for Help”).

Signs of Safety is a culturally and linguistically Deaf-accessible toolkit to be used alongside Seeking Safety (Anderson et al., 2021). The toolkit includes a supplemental therapist guide and population-specific client materials (e.g., visual handouts designed by a Deaf artist; ASL teaching narratives on digital video). It is designed for Deaf/signing clinicians, as well as non-signing clinicians working with ASL interpreters.

Control condition

Participants assigned to the control condition were asked to wait 16 weeks before receiving the 12-session Seeking Safety + Signs of Safety intervention from their assigned study therapist. At the time of study design, this 16-week waiting period was equivalent to the typical waitlist to receive psychotherapy from an ASL-fluent provider in Massachusetts. During this 16-week waiting period, control participants completed assessment interviews on the same schedule as participants in the experimental condition (described in Measures below).

Measures

Clinical outcome measures

Across study conditions, participants were assessed at five time points: baseline; week four; week eight; week 12 (immediate posttreatment); and week 16 (1-month follow-up). Primary clinical outcomes at immediate posttreatment and 1-month follow-up were past 30-day alcohol use frequency/quantity and past 30-day PTSD severity. Table 1 lists study measures by assessment time point.

Clinical outcome measures were administered at each time point as structured ASL interviews through HIPAA-compliant Zoom or videophone for the Deaf. The study interviewer was a Certified Deaf Interpreter who specialized in mental health interpreting and was a bilingual user of English and ASL.

Feasibility measures

In addition to the clinical outcome measures described above, we assessed study recruitment, retention, and other key aspects of feasibility (see Table 2).

Regarding therapist fidelity to the treatment protocol, all intervention sessions were recorded via Zoom or screen capture, with 25% of the sessions assessed on an ongoing basis by a trained Seeking Safety fidelity rater (i.e., three sessions per participant). To quickly correct fidelity concerns and capture fidelity throughout the treatment process, one recording was randomly selected from intervention sessions 1–4, one from 5–8, and one from 9–12. Sessions were rated using the Seeking Safety Adherence Scale (Najavits, 2003).

Exit interviews

At the end of the study, participants and study therapists were offered the opportunity to participate in a supplemental exit interview conducted by our team’s postdoctoral fellow (co-author Wilkins), a Deaf ASL-fluent clinical psychologist. These one-hour, semi-structured in-depth qualitative interviews were conducted directly in ASL. The interviews explored participants’ and study therapists’ satisfaction, dissatisfaction, and constructive feedback specific to the research methods used in the Signs of Safety pilot clinical trial.

The interviews were recorded via screen capture and analyzed by our postdoctoral fellow using our team’s linguistically- and socio-politically-appropriate method for conducting qualitative research with Deaf sign language users (Anderson et al., 2018). In this method, ASL qualitative data are directly coded in their original form, rather than undergoing translation and transcription into written English for the purposes of data analysis. This method leverages video technology and video-enabled qualitative analysis software to delay the translation of qualitative data from ASL to English until after data analysis and interpretation have already occurred. This intentional delay of translation into English allows our team to analyze the data in its original sign language form and therefore preserves the linguistic accuracy of the data throughout the analysis process (Temple & Young, 2004).

Applying this method, Dr. Wilkins analyzed the qualitative data using an inductive approach that applied two major techniques: (a) summative content analysis, in which the number of similar responses to questions were tallied, categorized, and described (Hsieh & Shannon, 2005); and (b) a summary of the answers to the questions outlined by Casey (Krueger, 1998). Such questions include: What are the participants saying? What are they feeling? What is really important? What are the themes? What ideas will be especially useful for intervention? Are there any comments said only once but deserve to be noted? Which quotes really give the essence of the conversation? Once these representative quotes were identified, our postdoctoral fellow translated them from ASL to English solely for the purposes of inclusion in the present article.

Results

Sixteen participants enrolled in the Signs of Safety pilot feasibility clinical trial—five Deaf men and 11 Deaf women. Mean age was 43.8 years (SD = 11.9). Four participants identified as Hispanic or Latino. One identified as Black/African American. Eleven identified as white, non-Hispanic/Latino.

Below, we synthesize emergent themes from these participants’ exit interviews with feasibility data and the real-life challenges that our team encountered when implementing the Signs of Safety pilot clinical trial. Particularly illustrative participant accounts and quotes were selected from the qualitative data and translated from ASL to written English to help demonstrate five major lessons learned.

Lesson one: expand your reach

Recruitment difficulties are one of the most basic challenges researchers face when conducting clinical trials. However, these challenges intensify for research in the Deaf community. The primary reason is that the population from which we draw is relatively small—recent estimates suggest that approximately one million Deaf and Hard of Hearing American adults use sign language (Mitchell & Young, 2022), roughly .3% of the U.S. population. Members of the Deaf community are dispersed across the country, with higher concentrations in select urban areas (e.g., Washington, DC; Rochester, NY; Austin, TX).

Another major limiting factor to recruitment is the Deaf community’s collective feeling of mistrust toward the medical community. There is a significant theoretical disconnect between the Deaf community and the academic medical community. Hearing healthcare providers and clinical researchers generally follow a medical model to “cure” or “fix” deafness (McKee et al., 2013), whereas most Deaf people do not want to be fixed, but rather to be respected as a cultural and linguistic minority group (Bauman, 2004; Ladd, 2003; Lane, 1992).

This disconnect has fueled a long history of unethical treatment of Deaf people in the healthcare system and research settings—for example, the highly unethical sterilization of Deaf people during the 1880–1950 eugenics movement (Lane, 2005). More recent missteps of healthcare providers and clinical researchers include failure to provide accommodations or ASL interpreters for treatment and study procedures; an overwhelming focus on treatments and research questions meant to “solve the problem of deafness;” and failure to explain and obtain informed consent in Deaf people’s primary language, ASL (Lane, 2005; McKee et al., 2013).

Combined, the relatively small size of the population and the understandable reluctance to participate in clinical research present significant challenges to the recruitment of Deaf study participants. If research is limited to a particular healthcare system or region, obtaining sufficient sample size is unlikely, especially if the research is focused on a specific clinical condition or comorbidity. As such, our first lesson taught us the critical importance of expanding study reach to a nationwide level to recruit a robust sample.

Expanding study reach has implications for other challenges to study operations, like small community dynamics. In the Signs of Safety pilot feasibility trial, we encountered several instances where our Massachusetts-based study participants had pre-existing relationships or conflicts of interest with our Massachusetts-based research team members. In each of these instances, the principal investigator assessed the nature of the dual relationship, transparently discussed the dilemma with the participant, and collaboratively worked with the participant to find a solution (e.g., asking the team member to recuse themselves for this participant, modifying study methods to further protect privacy, etc.). Although such dual relationships are still possible in a nationwide clinical trial, they are much less likely to occur and easier to resolve if the research team and study participants are drawn from the entirety of the United States instead of just one state or region.

In addition to expanding geographic reach, it is also critical to expand study reach to a diverse array of Deaf, Hard of Hearing, and DeafBlind individuals. There are many different language experiences and modalities of communication within the heterogeneous Deaf community. Potential participants may self-exclude themselves from joining a study because they assume they are not within the target study population or that the specific treatment intervention is not accessible to them. For example, one study participant shared that they were initially worried that they might not like the Signs of Safety intervention because they started learning ASL in their teenage years; however, after receiving the treatment they reported benefits from the visual handouts and ASL teaching story videos, despite not being a native ASL user. Given this possibility of self-exclusion, recruitment materials should clearly identify the study’s target population—in our case, Deaf ASL users, including those who learned ASL later in life or are bilingual users of other languages.

Lesson two: assess for language fluency

In clinical trials conducted within the general U.S. hearing population, fluency in the majority language (i.e., English) is often assumed and rarely assessed formally. In clinical research where English fluency is formally screened (usually by self-report), as many as 19% of studies intentionally exclude non-English users from participation (Muthukumar et al., 2021). Furthermore, excluding participation based on English language fluency may be more likely in psychotherapy clinical trials than physiological clinical trials. It has been previously demonstrated that when analyzing “clinical trials in four therapeutic areas (depression, diabetes, breast cancer, and prostate cancer), trials related to depression were the most likely to require English proficiency (52.24%)” (Muthukumar et al., 2021, p. 2). Translating these practices to the Deaf community suggests that psychotherapy researchers should formally assess Deaf individuals’ level of ASL proficiency prior to study enrollment.

In the Signs of Safety pilot feasibility trial, we sought to recruit individuals who self-identified as “Deaf ASL users.” When we selected this inclusion criterion, we failed to predict the number of individuals with minimal language skills who would present for the study and our team’s lack of preparation for formally assessing their ASL proficiency. The most prominent example of this challenge is that we enrolled and randomized two language dysfluent participants. Both individuals had experienced a lack of accessible language exposure throughout childhood (referred to as “language deprivation”; Glickman, 2007; Hall et al., 2017), were not proficient in ASL or any other language, and were unable to establish communication with any of our ASL-fluent study therapists. In both cases, we attempted to provide additional accommodations via Certified Deaf Interpreters. Although such accommodation may have proved effective during in-person therapy sessions, it was ultimately unsuccessful in establishing communication in the virtual teletherapy context. Because of these unresolved communication barriers, these two participants were unable to grasp the meaning or purpose of “therapy” and, therefore, were unable to truly consent to study participation. After consultation with the Institutional Review Board of record, we considered these individuals’ consent to be invalid, removed them from the study, and offered them more suitable resources outside of the study (e.g., independent living support services, advocacy support services, community-based support services).

These complicated experiences taught us the importance of formally assessing ASL fluency before participants enroll in the study. In our upcoming clinical trial, we’ve included two eligibility screening questions that ask potential participants to self-report their level of fluency in ASL and written English literacy. Our Deaf Community Advisors will serve as a means of validating this self-report during the informed consent session, during which they will spend ~1 hour in interaction with the potential participant. Any concerns about potential participants not meeting minimum ASL fluency for the study will immediately be brought to the full research team to discuss collaboratively.

In addition to procedures that help researchers gauge a potential participant’s level of sign language fluency, it may also be useful to include screening questions that are potential markers of childhood language access versus childhood language deprivation. For example, did the individual have any Deaf family members, or were they the only Deaf person in their family? What languages were used in their childhood home? At what age did they begin learning ASL?

Another reason for assessing language fluency is to ensure a better participant-therapist language match. Three exit interview participants discussed the importance of making sure the therapist’s signing experience and skill level were a good match for the participant. Any language incompatibilities became a barrier to effectively participating in treatment. As such, a formal assessment of therapists’ language skills is also necessary to ensure a good language match between study participants and their assigned study therapists. One option for researchers who wish to formally evaluate the language skills of study therapists could be the ASL Proficiency Interview (ASLPI), a holistic language evaluation conducted by the Bilingual Evaluation, Testing, and Assessment Center at Gallaudet University. The ASLPI is used to determine an individual’s global ASL proficiency via a face-to-face interview that tests what an individual can do with the target language at a given point in time. Although such procedures do not guarantee effective communication between Deaf study participants and ASL-fluent study therapists, they increase the likelihood that effective communication will be established.

Lesson three: don’t manufacture a waitlist

Another challenge researchers often encounter when designing psychotherapy clinical trials is selecting an appropriate control condition. One commonly used approach is the waitlist control condition, in which study participants might be randomized to an artificial waitlist. For these participants, treatment is withheld until all study-related outcome assessments are complete. The wait periods for such studies vary based on the duration of the treatment protocol, as well as the duration of the posttreatment follow-up period.

In the Signs of Safety pilot feasibility trial, we randomized participants to immediate treatment or to a waitlist control condition, in which they were asked to wait to receive the experimental therapy until 4 months later (12-week treatment protocol + 1-month follow-up period). Despite its standard use in clinical trials research, the decision to implement a waitlist control condition was very difficult for our team. It weighed heavily on us, especially members of the team who were Deaf community members themselves. Our team is all too familiar with the multitude of barriers that Deaf people experience to accessing mental health services (Cabral et al., 2013; James et al., 2022a; McDonnall et al., 2017; National Association of the Deaf, 2003; Steinberg et al., 1998). As such, withholding treatment from individuals with active trauma and alcohol problems, even if it was only for a period of 4 months, felt cruel and unnatural.

Our feasibility findings validated these concerns. Seventeen percent of eligible participants chose not to enroll in the study because they were highly motivated for treatment at the time of screening, needed immediate treatment, and would not risk being placed on a 4-month waitlist. Others begrudgingly consented, knowing they had few options to access treatment outside the study. Yet, of the enrolled participants assigned to the waitlist condition, only 57% took advantage of the 12 free treatment sessions after the 4-month waiting period, suggesting that we missed the critical opportunity to intervene at a transitory moment of readiness for change. Furthermore, only one waitlist participant was willing to participate in the exit interview process. They reported that if they were not attending daily peer recovery groups during the waiting period, they would have “relapsed for sure.”

“For ethical reasons, waiting list controls have been preferred to no treatment controls, provided the wait is still shorter than that for routine services. However, could significant differences arise from the wait being detrimental rather than the intervention being beneficial?” (Elliott & Brown, 2002, p. 1047). Waitlist control conditions are considered ethnically sound when the study wait period is shorter than the typical wait for similar services. Yet, how does this reasoning apply when Deaf-accessible therapy services are either completely unavailable or require an extremely long wait to begin treatment? Does the waitlist control condition unintentionally create a coercive situation in which the individual has no choice but to join the clinical trial in order to access treatment? In hindsight, our team regretted implementing a waitlist and asking participants to delay treatment receipt, especially after learning that 43% of our waitlist participants never sought their therapy sessions following the waiting period.

Our team has chosen a different approach for our upcoming full-scale clinical trial of Signs of Safety, to be conducted in collaboration with the Deaf-owned agency National Deaf Therapy (NDT). Ninety-six individuals from the U.S. states served by NDT will be randomized to receive experimental therapy or therapy as usual (i.e., basic supportive therapy). Both active treatment conditions will be provided by Deaf clinicians who are fluent in ASL. To address the need for a control condition, we will enroll an additional 48 individuals into a third condition—a contemporaneous no-treatment control arm. These individuals will be recruited from the existing NDT waitlist, comprised of Deaf individuals residing in the states not yet served by NDT but voluntarily awaiting NDT services. In other words, we will use NDT’s natural waitlist to feed the proposed control condition. Data from control participants who gain access to treatment during the study protocol (i.e., come off NDT’s waitlist or gain access to formal psychotherapy elsewhere) will be kept and analyzed separately from the main outcome data.

Given that our study protocol is 9 months in length (3-month treatment period + 6-month posttreatment follow-up period), it would be ethically unfathomable to create an artificial waitlist and ask participants to refrain from desperately needed treatment, especially given that our study population remains severely underserved across the U.S. behavioral healthcare system. Moreover, our planned approach aligns with concerns that a manufactured waitlist condition may artificially inflate intervention effect estimates—participants placed on the waitlist when actively seeking treatment may mistakenly believe that they are being asked to wait before self-initiating any positive behavioral changes (Cunningham et al., 2013).

Lesson four: streamline video fidelity monitoring

The field of psychotherapy research has primarily relied on audio recordings of treatment sessions to monitor therapist fidelity to the intervention protocol. For therapy sessions conducted in spoken language, this approach makes a lot of sense—audio recording is relatively unobtrusive, cost-effective, and protects privacy by only capturing the voices of those recorded. To capture therapy sessions conducted in a visual modality—sign language—determining an effective and efficient approach to fidelity monitoring via video recording was one of our most challenging endeavors during the Signs of Safety pilot feasibility trial.

The first obstacle was the obtrusiveness of video recording. Prior to the onset of the COVID-19 pandemic and the shift to video telehealth, we asked study therapists to set up a camcorder in their therapy office and situate themselves close enough to the participant so that both the therapist and participant would be captured in the recording. This approach proved to be physically and psychologically uncomfortable for everyone involved. One participant who experienced both in-person and virtual sessions said they “preferred in-person sessions because of the energy,” but acknowledged that the video camera was obtrusive and required the therapist and participant to sit closer together than they normally would have. Two participants recommended less intrusive video methods for in-person sessions, such as recording on a smartphone or small camera, and having the camera set up before participants arrive. Similar findings have been reported when video recording is used for fidelity monitoring with hearing clinical trial participants, noting that videotaping is “more obtrusive and costly” and “may increase demand characteristics” (Borrelli, 2011, Table 1), the likelihood that the participant will change their behavior in response to being recorded.

In some ways, the obtrusiveness of recording partially resolved itself with the need to pivot to video telehealth in March 2020. Regardless of whether participants received treatment in person or via Zoom, they provided express consent to be video recorded—this consent was obtained both during the informed consent process and at the beginning of each session. However, being recorded via the Zoom recording feature is much less noticeable to the participant than a blinking camcorder pointed in their direction. This therefore reduces demand characteristics throughout the course of the therapy session. The increased comfort for participants, as well as the ease with which our study therapists were able to record their sessions for fidelity monitoring, were two important factors that led us to design our upcoming clinical trial in the virtual therapy realm.

However, one exit interview participant expressed strong dislike of the recording process, even when conducted virtually: “It felt weird, the recording felt weird…I would prefer no recording at all. It made me feel like Big Brother was watching. I don’t know if the people watching were going to understand my life story and the things I was sharing. Maybe it would be better if they only recorded a few sessions, like the first one, one in the middle, and then the last session…I hated that red dot staring at me, like an eye.” The participant also shared that they didn’t remember the purpose of the video recording and wished that they had been more assertive in asking why their sessions were being recorded.

While the purpose of recording is discussed in depth during the informed consent process, participants would benefit from reminders about the purpose of recording during the first therapy session and occasionally throughout the treatment course. This would remove the burden of the participant needing to ask or wonder about the purpose of recording. In addition, recording methods that show no indication that recording is taking place, such as screen video capture applications, could be a good alternative for those participants who feel bothered by a visual indicator that recording is taking place. Screen capture recordings are also less prone to video capture errors that may occur in Zoom due to the type of view being used during the meeting, the use of multiple monitors, and the use of screen sharing features.

Our second obstacle to video fidelity monitoring remains to be fully resolved—the large size of video files compared to audio files and how to securely transmit these large files from study therapists to the research team. Fortunately, our institution supports an approved secure file transfer system that can transmit large video files. Unfortunately, to transmit one 45-minute therapy session requires hours of upload time on the part of the study therapist and hours of download time for the research team. We are hopeful that ongoing technological advances will eventually reduce this burden. In the meantime, our approach for our upcoming clinical trial will be for study therapists to transfer video files for only a small, random sample of their sessions—a suggestion that aligns with participant exit interview data. We will also compensate study therapists for the time needed to upload recordings at a pay rate equivalent to their time spent providing therapy services to prevent loss of wages due to study procedures.

Lesson five: make outcome assessment accessible

Our fifth and final lesson also relates to language—not surprising given the critical value placed on ASL and language access within the U.S. Deaf Community. This lesson may seem obvious—conduct assessments in the participant’s primary language—however, only 2.7% of clinical trial protocols report that they offer translations into languages other than English (Muthukumar et al., 2021). U.S. academia is an English-majority environment, and many researchers are primary English users. As a result, clinical trial research methods are skewed towards the primary language of English-speaking researchers’ primary language, rather than participants’ language access needs—“methodological expediency continues to reinforce the political invisibility of the language and its users” (Temple & Young, 2004, p. 166).

Specific to the Deaf ASL users, most hearing researchers assume that Deaf study participants are fluent in written English and that, in the absence of ASL, written English can be relied upon as an alternate means of communication. This is true of some Deaf people, but not of others. Due to a range of medical and educational barriers, English literacy is more variable in the Deaf population than in the general hearing population (Morere, 2011).

In the Signs of Safety pilot feasibility trial, our team attempted to address these language access considerations by conducting outcome assessments via live interviews instead of written English self-report surveys. As described in the Method section, we hired a Certified Deaf Interpreter who specialized in mental health interpreting to conduct structured interviews with participants, directly in ASL. However, this approach elicited additional challenges that we did not originally anticipate.

First, participants reported mixed feelings about the live interview format, especially given the sensitive nature of some of the questions being asked (e.g., questions regarding illicit drug use, sexual symptoms related to trauma, etc.). “Some of the questions were repetitive and as someone who was newly sober, as is common for addicts, I was very suspicious. I was like, is this a test? Are they testing me if I answer wrong or differently on the two similar questions? That was only at the beginning though. I remember by the last few interviews, my feelings, my perspectives had changed. I loved watching myself change and grow and one of the ways was that I became less paranoid and suspicious.” Some study participants later disclosed that their responses during the interviews were not completely authentic due to demand characteristics, embarrassment, feelings of shame, or fear of judgment. They indicated that they would have preferred to respond to a self-guided survey rather than respond to a live interviewer. “Talking about trauma was easier because I’ve been in therapy for a long time. But, talking about addiction…that part was a little harder because, I don’t know, I’ve never really discussed it with anyone and that’s because of trust. The interviewer seemed safe. She was very careful. My difficulty wasn’t her asking the questions but rather the topic itself was very difficult for me.”

Second, on a practical level, scheduling and coordinating five longitudinal interviews between each participant and the study interviewer was a logistical challenge. This administrative burden was manageable for a small pilot study but is not scalable for a nationwide clinical trial with more than 100 participants.

Third, an interview approach may be less standardized than a crystallized questionnaire, even when following a structured interview format as in the Signs of Safety pilot clinical trial. Even equipped with a structured interview protocol, a live interviewer may deviate from the predetermined structure. In fact, this is just what Certified Deaf Interpreters are trained to do—find alternate ways to communicate complex concepts to match a wide range of language registers of their Deaf consumers. This is a constant area of discussion in Deaf clinical research—should we prioritize rigid standardization, or should we prioritize matching the language access needs of each unique participant? Our current work attempts to satisfy these competing demands.

To prepare for the Signs of Safety full-scale clinical trial, our team undertook a Deaf community-engaged process to translate 13 validated, open-access measures of PTSD, alcohol use, drug use, and mental health from written English into ASL. Of these measures, only one had previously been translated into ASL. The final translations were professionally filmed and will be uploaded into an ASL-accessible online survey platform. We anticipate that this approach will be responsive to participant’s privacy concerns, reduce administrative burden on the research team, and resolve concerns regarding standardization of our assessment protocol. Once we begin to amass participant responses and can formally assess the psychometric properties of these ASL translations, our goal is to create an open-access repository so that these new measures can be used widely across the field of Deaf mental health.

Conclusion

Above, we describe our methodological lessons learned from a psychotherapy pilot clinical trial conducted in the Deaf community between 2019 and 2022. Compared to therapy clinical trials conducted in the general hearing population, studies with Deaf sign language users require unique considerations regarding the scope of recruitment, design of eligibility screening criteria, selection of control conditions, adaptation of fidelity monitoring procedures, and translation of study assessment measures. These methodological adaptions revolve around three key characteristics of the U.S. Deaf population—(a) a small, close-knit community that (b) highly values access to ASL and, despite a marked need for support, (c) continues to be severely underserved by the U.S. behavioral healthcare system.

These five lessons learned were crucial in our design of the first-ever full-scale psychotherapy trial in the U.S. Deaf community, which will be conducted from late autumn through 2028 (NCT06278922). However, these design considerations are merely a starting place. As we conduct our full-scale therapy trial across the coming years, we recognize that more problems will arise and require creative solutions. Our team will diligently document these problems and solutions so that, at the end of the trial, we can disseminate a clear roadmap to guide researchers in the effective conduct of therapy clinical trials with Deaf sign language users.

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: M. Anderson, A. Wilkins, S. Hostovsky, E. Pici-D’Ottavio, A. Aldalur; data collection: M. Anderson, A. Wilkins, E. Pici-D’Ottavio; analysis and interpretation of results: M. Anderson, A. Wilkins; draft manuscript preparation: M. Anderson, A. Wilkins, A. Aldalur, F. McGinnis, K. Meza. All authors reviewed the results and approved the final version of the manuscript.

Funding

This work was supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) of the National Institutes of Health (NIH) [grant numbers R34AA026929, R34AA026929-S, and R01AA031010]. The content is solely the responsibility of the presenter and does not necessarily represent the official views of NIH.

Acknowledgments

We would like to express our deep gratitude for the ongoing research guidance provided by Drs. Lisa Najavits, Bruce Barton, and Beth Epstein. To the Signs of Safety pilot clinical trial therapists, a huge “thank you” for your long-lasting commitment to this work—Ms. Toni Butland, Ms. Justine Stohl, Dr. Karen Regan, and Dr. Neil Glickman. Last but not least, THANK YOU to all the Deaf community members and Signs of Safety research participants who have contributed to this research—we could not do it without you!

Ethical Approval

The research described in this article was approved by the UMass Chan Medical School Institutional Review Board, protocol numbers H00016364 and STUDY00001149.

Competing Interests

The authors report no relevant financial or non-financial competing interests.

Clinical Trial Registration

The Signs of Safety pilot feasibility trial is registered on ClinicalTrials.gov with ID number NCT03845985. The Signs of Safety full-scale clinical trial is registered on ClinicalTrials.gov with ID number NCT06278922.

References