key: cord-0259393-my1235wp
authors: Edgar, N. E.; Bennett, A.; Dunn, N. S.; MacLean, S. E.; Hatcher, S.
title: Feasibility and acceptability of Narrative Exposure Therapy to treat individuals with PTSD who are homeless or vulnerably housed: A pilot randomized controlled trial.
date: 2021-11-11
journal: nan
DOI: 10.1101/2021.11.08.21266074
sha: 53b2ff5751f603d6f34ee2d72ec05f21fa0f83a6
doc_id: 259393
cord_uid: my1235wp

Background: Annually, there are least 235,000 individuals experiencing homelessness in Canada. These individuals are more likely to have complex health issues, including mental health issues such as post-traumatic stress disorder (PTSD). Diagnosed PTSD rates in the homeless are more than double that of the general population, ranging between 21% and 53%. In the homeless population, complex PTSD (cPTSD) appears to be more common than PTSD. One treatment option for cPTSD is Narrative Exposure Therapy (NET), a brief trauma focused psychotherapy which attempts to place the trauma within a narrative of the person's life. Previous studies suggest NET may be an effective option for those who are homeless. In this study, our primary aim was to assess the feasibility and acceptability of delivering community-based NET to individuals with PTSD who were homeless or vulnerably housed. Methods: This pilot randomized controlled trial (RCT) enrolled participants who were 18 years of age or older, currently homeless or vulnerably housed, and with active symptoms of PTSD. Participants were randomized to either NET alone or NET plus the addition of a genealogical assessment. Demographic and clinical data were collected at the baseline visit. Symptoms of PTSD, drug use and housing status were re-assessed at follow-up visits. Rates of referral, consent and retention were also examined as part of feasibility. Results: Twenty-two potential participants were referred to the study. Six were not able to be contacted, one was excluded prior to contact, and the remaining 15 consented to participate. Of these, one was a screen failure and 14 were randomized equally to the treatment arms. One randomized participant was withdrawn for safety. The main point of attrition was prior to starting therapy (3/13). Once therapy was initiated, retention was high with 80% of participants completing all six sessions of therapy. Seven participants completed all follow-up sessions. Conclusion: Delivering NET in a community-based setting and completing genealogical assessments was both feasible and acceptable to those who are homeless or vulnerably housed. Once therapy had been initiated, participants were likely to stay engaged. A large RCT should be conducted to evaluate effectiveness and feasibility on an increased scale.

Homelessness is a rapidly growing problem in Canada with at least 235,000 individuals 107 experiencing homelessness every year(1,2). In reality this number is likely 3 to 4 times higher to 108 account for individuals with no real prospect of permanent housing options or "hidden 109 homeless"(1). In Canada, point in time counts have repeatedly shown year over year increases 110 in homelessness(2). Exacerbated by a nationwide housing crisis and a global pandemic, 111 accompanied by a reduction in shelter beds and social services, 2021 reports have shown that 112 the situation is continuing to worsen(3-5). 113

Individuals experiencing homelessness are more likely to have poor overall physical and mental 114 health, increased mortality rates and to experience more barriers to accessing healthcare(6,7) 115 compared to their housed counterparts. In one study nearly 50% of respondents indicated 116 having 3 or more physical health conditions, and 52% indicating having a mental health 117 diagnosis(6,8). Individuals who are homeless are more likely to be exposed to infectious 118 diseases, such as tuberculosis or hepatitis A (6); less likely to seek early treatment and less 119 likely to receive care equivalent to those who are housed (9-12); and face additional challenges 120 with adherence to treatment regimens (13) . 121

Exposure to trauma is a nearly universal experience among the vulnerably housed. It is 123 estimated that as many as 91% of individuals who are homeless have experienced at least one 124 traumatic event(14) and up to 99% have experienced childhood trauma(15,16) Diagnosed post-125 traumatic stress disorder (PTSD) rates in the homeless are significantly higher than the 126 Canadian population, ranging between 21% and 53%(17-20), compared with a lifetime 127 prevalence of 9.2%(21) in the general population. In addition to trauma before becoming 128 homeless, the experience of being homeless increases the risk of exposure to traumatic events 129 (22). Many who are homeless are exposed to violence, with previous work showing 40% of 130 individuals reporting being assaulted and 21% of women reporting being raped in the previous 131

year (6). Lastly, the experience of being homeless is itself traumatic due to the loss of shelter, 132 safety, stability, and, often, social supports. Thus, being homeless continues to re-traumatize 133 and victimize the individual(23). 134

In ICD-11, a distinction is made between PTSD and complex PTSD (cPTSD). cPTSD is 135 characterised by experiencing trauma that is prolonged or repetitive from which escape is 136 difficult or impossible (for example, repeated childhood sexual or physical abuse)(24). This 137 results in the symptoms of PTSD plus problems in affect regulation, negative self-beliefs and 138 difficulty sustaining relationships. In the homeless population cPTSD appears to be more 139 common than PTSD, with one survey of 206 homeless adults finding 60% diagnosed with 140 cPTSD and 16% with PTSD(25, 26) . 141

This subsequently impacts how individuals engage in health care services, with individuals often 142 experiencing distrust of both people, including healthcare providers, and services(27). It may 143 also lead to self-medication with street drugs to address the symptoms of PTSD. There are also 144 systemic barriers to accessing care which include difficulties finding transportation to 145 appointments, institutional rules that effectively ban people who are homeless, feelings of 146 stigmatization(9,11), having proof of health insurance, and access to no-cost mental health 147 services(28). Most recently, the shift to primarily virtual mental healthcare has further isolated 148 this population from accessing services. Providing therapy in this population should consider the 149 challenges of structural exclusion that this population faces with respect to healthcare services 150 and providers, as well as the unique symptoms associated with cPTSD. 151

Narrative Exposure Therapy 152 Narrative Exposure Therapy (NET) is a brief trauma focused psychotherapy and was developed 153 based on principles derived from exposure therapy, cognitive behaviour therapy and testimony 154 therapy(29). NET attempts to place the trauma within a narrative of the person's life. This 155 therapy has been evaluated in traumatized populations with a focus on survivors of conflict and 156 organized violence(29). NET is recommended for treatment of PTSD in several guidelines, such 157 as the American Psychological Association guidelines (30) and the National Institute for Health 158 and Care Excellence guidelines(31). There are three therapeutic components which consist of 159 education about the effects of trauma, constructing a biography and narration of traumatic 160 events. The autobiography is recorded by the therapist and is built upon with each subsequent 161 reading. A focus of the therapy is to integrate the generally fragmented reports of traumatic 162 experience into a coherent narrative and to bring about the habituation of emotional responses 163 to reminders of the traumatic event(29). There have been no trials of NET in homeless adults, 164 although one study of NET with 32 street children found a reduction in self-reported 165 offences(32). Anecdotal evidence of using this approach in the homeless population suggests 166 that constructing an autobiography helps to give meaning to problems and provides the initial 167 steps in constructing a core sense of belonging and identity. There is also some evidence that 168 NET may have advantages in treating complex traumatization seen in disadvantaged 169 populations compared to typical first line therapy, such as Prolonged Exposure Therapy(33). 170

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ;  https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint Genealogy has been used in family therapy(34) and counseling(35) to promote identity(36) and 172 develop connections to ancestors. This can improve relationships with living relatives and 173 potentially address the negative self-beliefs which are part of cPTSD. A systematic review of the 174 acceptability of health and social interventions for people who were homeless found that having 175 a positive self-identity improved links to services(27). The link to those who have gone before is 176 a common theme in indigenous health(37). Previous experience of using problem solving 177 therapy, with a focus on a sense of belonging, in Māori in New Zealand who had presented to 178 hospital with intentional self-harm resulted in improved outcomes after a year compared to usual 179 care. The focus on sense of belonging helped to re-frame individuals' narrative beyond the 180 immediate family. The metaphor used by participants was that knowing about previous 181 generations helped to deepen their roots so they were less likely to be blown over by life's 182 storms(38). 183

As of February 2021, there were no randomized controlled trials of trauma focused therapies in 185 people who are homeless with PTSD. The aim of this study is to test the feasibility and 186 acceptability of delivering community-based NET to individuals with PTSD who were homeless 187 or vulnerably housed. We also extended the option of genealogist support to evaluate the 188 potential impact of this experience on the development of their narrative and self-identity. This 189 trial forms part of the preparation phase of a multi-phase optimization strategy (MOST) for 190 developing and delivering treatment for PTSD in people who are homeless using trauma 191 informed care. Using the outcomes from this study, we will design a full-scale RCT to evaluate 192 and optimize a model of trauma-informed care incorporating NET for the treatment of PTSD in 193 the homeless. 194

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint is a community-based short-term service providing support and referrals to individuals who are 207 homeless or vulnerably housed experiencing severe and persistent mental illness. Eligibility 208 criteria are described in Figure 1 . 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint

Participants were randomized to one of two groups: Narrative Exposure Therapy only (NET) or 213

Narrative Exposure Therapy supplemented with a genealogist assessment (NET+G Sessions 3 -5 Narration of the lifeline from birth through each event • Traumatic events are confronted and reprocessed until arousal response decreases • At follow-up sessions, the draft narrative is read through collaboratively, focusing on important events.

The narrative is updated and corrected, providing more clarity with each read-through • This is repeated until a final version of the narrative is completed (by Session 6) Session 6

• The final narrative is read through entirely and signed by the participant and therapist.

The participant is provided a copy of their narrative 

Participants were met in the community at a mutually agreed upon location. Visit locations 227 included shelters, outreach offices, community day programs and our research office. Locations 228 were chosen to minimize participant burden and increase feelings of comfort and safety. Any 229 required travel costs were covered by the study. 230 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint After referral to the study, participants met with a trained research assistant to complete the 231 informed consent process and screening procedures. Screening procedures to confirm eligibility 232 involved a structured interview completed by the research assistant to confirm active symptoms 233 of PTSD and housing status. Once eligibility was confirmed, participants completed 234 assessments collecting demographics and evaluating general mental health, alcohol and 235

substance use, quality of life, health care utilization, and cognitive state. We collected 236 sociodemographic information on gender (male, female, transgender, non-binary), self-identified 237 ethnicity, highest level of education completed, marital status and medical history. 238

The primary outcomes of this study were the feasibility and acceptability of a completing an 240 RCT of NET in the homeless population. The primary measure of feasibility was recruiting the 241 planned sample size over six months. In addition, we wanted to determine the acceptability of 242 NET in this population which we defined as at least 50% of those approached about the study 243 consenting to take part. Lastly, we wanted to see if it was feasible to collect outcome data in this 244 population which we defined as restricting study dropouts or lost to follow ups to 25% of 245

We also wanted to see if NET treatment in this population resulted in better health-related 247 outcomes including: a decrease in the severity of PTSD symptoms, change in housing status, 248 improved overall health, better health-related quality of life and lower rates of alcohol/drug 249 misuse. Other secondary outcomes included creating a training manual for Narrative Exposure 250

Therapy in this population that also included the incorporation of a genealogist. 251

Measures and timing of administration are outlined in Table 2 . Housing status was collected 252

prior to entry into the study and at each follow-up visit. Participants self-identified their housing 253 among the following options: living in a shelter, living with a friend, living with family, 254 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

AUDIT A brief 10-item self report questionnaire assessing alcohol misuse X -X X X ADHD Self-Report Scale

An 18-item self-report questionnaire that assesses the inattentive (6 items) and hyperactive/impulsive (12 items) dimensions of ADHD

A brief interview assessing multiple cognitive domains including visuo-constructional skills; naming; memory; attention; sentence repetition; verbal fluency; abstraction; delayed recall; and, orientation.

The SF-20 was used from February 2019 -September 2019. This is a 20-item questionnaire that assesses various health outcomes, and the extent to which health related problems interfere with daily life.

X -X X X . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint

The EQ-5D-5L was used from September 2019end of study. The EQ-5D-5L is 5-item questionnaire that assesses health-related quality of life.

A self-report questionnaire asking about missed time from work/volunteer/school and utilization of healthcare services.

We planned to enroll 12 participants in each arm for a total sample of 24 participants(41). The 268 study was not powered to test efficacy outcomes. 269

Randomization was completed by the Ottawa Methods Centre at the Ottawa Hospital Research 271

Institute (OHRI) with allocations kept in sequential sealed opaque envelopes at the OHRI study 272

office. Participants were randomized in a 1:1 allocation with no restrictions. After providing 273 consent and confirming eligibility, participants were randomized by a trained research assistant 274 according to the allocation in the sealed envelope. 275

All statistical analyses were conducted using IBM SPSS 26. Non-parametric data were 277 described using frequencies and percentages. Continuous data were described using measures 278 of central tendency (mean and standard deviation) and bivariate relationships were explored 279 using independent samples t-tests (t) and analyses of variance (ANOVA). Paired samples t-280

tests were conducted to explore any within subject relationships for PCL-5 score and substance 281 use patterns. Bivariate relationships were detected using two-sided Fisher's Exact Test (2x2 282 contingency tables), except for one-sided tests where noted in the tables. 283

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint Qualitative interviews were planned with participants who completed the study and service 285 providers including shelter, outreach, and day program staff. The service provider interview 286 evaluated their participation in the study, their perceived need for community-based NET to be 287 Results 296

Participant flow is outlined in Figure 2 . Recruitment for this study took place between February 298 2019 and February 2020 and a total of 22 people were referred to the study, (one of these was 299 immediately before the first pandemic lockdown in late February 2020). Recruitment was 300 stopped on two occasions, for a total of three months, as therapists had reached their maximum 301 case load. Five participants were referred to the study after the start of the COVID-19 pandemic 302

in March 2020, but we were unable to recruit them as there were no safe places to meet face to 303 face as most drop-in centres or other spaces were closed because of COVID-19. We have not 304 included them in the tables. Virtual appointments were not an option for this population due to a 305 lack of technology, access to the internet, or safe and private places to conduct therapy. 306 307 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The clinical teams referred 22 people to take part in the study over the nine months that the 312 study was accepting potential participants. One potential participant was deemed ineligible by 313 the Principal Investigator due to a significant brain injury and associated psychosis. Six (6/22, 314 27%) potential participants could not be contacted to arrange a baseline visit and consent, 315 leaving 15/22 (68%) who were assessed for eligibility for the study. One of these potential 316 participants did not meet the inclusion criteria. This participant consented but was unable to 317 complete the screening procedures. This led the team to evaluate the administration of 318 screening measures to reduce burden on participants. 319

This left 14 participants who were randomized. Seven were randomised to NET and seven to 320 NET+G. One participant who consented to the study and was randomized to the NET alone arm 321 did not complete their baseline assessment. While consent, screening and baseline typically 322 occurred at one visit, this participant had a particularly long screening session, and the baseline 323 assessments were scheduled to occur during a second encounter. The participant did not 324 present for this visit and further information from the referrer showed a possible safety risk to the 325 therapists, so this potential participant was withdrawn. This participant had no evaluable data 326

and was not included in the evaluation of acceptability or effectiveness. 327

Three people were lost to follow-up between consenting to take part in the study and the first 328 therapy session, one in the NET+G group and two in the NET group. 329

Every effort was made to provide participants with both their personal narrative and genealogy 330 report before their final NET session. In particular, delays with genetic matching meant the first 331 participant in the NET+G arm did not receive their report until after their final session. 332

Adjustments were made in the timing of the genealogy interview so that all subsequent 333 participants received their reports prior to completing therapy, and typically around week 4. 334

Feasibility 335 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint

We were unable to recruit our desired sample size within the goal of 6 months. The enrollment 336 of 15 participants took place over approximately 9 months. This was primarily due to a shortage 337 of trained therapists to take on participants. For the first 5 months, only one therapist was 338 involved in the study. As the study progressed, we were able to recruit one additional trained 339

Study Acceptability 341 Figure 2 outlines the retention rates at various stages of the study. While we did not meet the a 342 priori threshold of 75% retention at week 12, 61.5% (8/13) of participants completed at least one 343 post-therapy follow-up assessment visit (week 8). Of the ten who started therapy, seven 344 completed all study visits up to week 12. Of three participants who did not complete the week 12 345 assessment, the first completed the week 8 study visit and was interested in completing week 346 12, however, the final visit was rescheduled due to personal circumstances and was eventually 347 cancelled due to the COVID-19 pandemic; the second dropped out at week 6 as they found the 348 sessions too triggering and difficult to complete while sleeping rough; and, the third individual 349 was lost to follow up at week 4 but re-engaged with the clinical team at a later date. 350

Of the 10 participants who started therapy, eight (80%) completed all 6 sessions. All participants 352 who were randomized to the NET+G arm (n=6) accepted the referral to complete their family 353

Demographics 355

The demographics of participants are outlined in Table 3 . Comparable to community 356 demographics, our sample contained more males than females (61.5%, 8/13), with a lifetime 357 history of drug use (84.6% ,11/13) and alcohol use (84.6% ,11/13). The sample self-identified 358 primarily as white (65%, 8/13), but also First Nations (7.7%, 1/13), Métis (2/13, 15.4%), Asian 359 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint (7.7%, 1/13), and Other (Hispanic) (7.7%, 1/13). Within this identification, two people also 360 identified themselves as mixed-ethnicity (Métis/White and First Nations/Black). Educational 361 background was diverse with 38.5% (5/13) having less than high school, 30.8% (4/13) having a 362 high school diploma or equivalent, 15.4% (2/13) with a college or university education, and 363 15.4% (2/13) with a graduate or professional degree. Marital status was also varied, with 46.2% 364 (6/13) identifying as single, 7.7% (1/13) as married, 23.1% (3/13) as separated, 7.7% (1/13) as 365 divorced and 15.4% (2/13) as widowed. No individuals identified as transgender or non-binary. 366

No participants indicated common-law status. 367 

All participants had experienced multiple traumatic events in their lifetime, with childhood trauma 371 being common, as reported on the Life Events Checklist. With respect to symptom severity, 372 baseline PCL-5 scores did not differ (F(1,13) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 11, 2021. ;  https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint PTSD scores decreased in both groups over the course of the study. Assessing the change in 375 symptom severity within subjects for the 7 who completed post-therapy follow-up to week 12, 376 showed a statistically significant improvement in PTSD scores (Table 4 ). Prior to initiating 377 therapy, the average PCL-5 score was 64.14 (SD 8.80). At the 12 week follow-up, participants 378

reported an average decrease of 17.29 points (SD 6.18), for a total PCL-5 score of 46.86 (SD 379 16.63) (t=2.798, df=6, p=.031, d=1.057) . This reduction in severity within subjects is also 380 considered a clinically meaningful change, defined as a reduction in total score by 10-20 381 points(42). 382 

A lifetime history of substance use was common in this sample, with 84.6% (11/13) reporting a 386 history of drug or alcohol misuse. At enrolment, 6 individuals reported current drug use (46.2%, 387 6/13) and 2 reported alcohol use (15.4%, 2/13) in the 30 days prior, while 5 participants reported 388 no drug or alcohol use at all during this period (38.5%, 5/13). Alcohol or drug use did not change 389 over the duration of the study (Table 4 ). Prior to initiating therapy, drug use was an average of 390 11.57 days (SD 14.47 days), while alcohol use averaged 1 day (SD 1.73 days). At week 12, 391 there was no significant decrease in drug use (t=1.00, df= 6, p=.356) or alcohol use (t=1.08, 392 df=6, p=.321). 393

Substance use appeared to have no relationship to whether or not a participant completed the 394 study, with 23.1% of participants using drugs at enrolment completing the study (p=.617, 395 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 11, 2021. ;  https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint Fisher's Exact Test, 1-tailed) compared to 30.7% of non-users completing the study. 396

Participants who did not complete the study were split, with 23.1% using drugs at enrolment and 397 23.1% not using drugs (Table 5) . 398 

Similarly, no statistically significant difference was found between those who completed the 401 study and were using alcohol (2/13, 15.4%) or not using alcohol (5/13, 38.5%) compared to the 402 six participants not using alcohol (6/13, 46.2%) who did not complete the study (p=.269, Fisher's 403 Exact Test, 1-tailed) ( Table 5) . 404

Housing status over the duration of the study is described in Table 6 . At baseline, more than 406 half the participants were living in a shelter (53.8%, 7/13), followed by supportive/transitional 407 housing (15.4%, 2/13), while 1 individual was living with a friend (7.7%), and another was 408 paying for a living space (7.7%). Several individuals added specifiers to their situation including 409 that they were paying for a living space because they received a subsidy or assistance from a 410 family member. During the study, one individual (7.7%) left the shelter and was not able to find 411 other housing options. Of those who did not complete the study, 38.5% (5/13) resided in a 412

shelter baseline (p=.078, Fisher's Exact Test, 1-tailed). 413 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint With Family 1 (7.7) 1 (7.7) 1 (7.7) 1 (7.7)

With Friend 0 (0) 0 (0) 0 () 1 (7.7)

Supportive/Transitional Housing 4 (30.8) 2 (15.4) 1 (7.7) 1 (7.7)

Paying for a Space 1 (7.7) 1 (7.7) 2 (15.4) 1 (7.7)

No Housing Options 0 (0) 1 (7.7) 0 (0) 0 (0) Missing 0 (0) 1 (7.7) 0 (0) 0 (0) Withdrawn 0 (0) 4 (30.8) 5 (38.5) 6 (46.2)

Of those who completed the study, 3 individuals remained at the shelter (3/7), while others were 416 staying with friends (1/7), staying with family (1/7), living in supportive housing (1/7), or paying 417 for a space (1/7). Several individuals experienced changes to their housing situation: two moved 418 from supportive living to the shelter, one from the shelter to living with a friend, one from 419 supportive housing to paying for a space, one from living with family to paying for a living space. 420

Of the 6 individuals who did not complete the study, 4 were at the shelter, 1 was "sleeping 421 rough", and 1 was paying for a living space. 422

We were unable to complete qualitative interviews with the participants because of COVID-19 424

restrictions. There were no safe places to conduct interviews and using technology was not 425 feasible as discussed above. Four service providers completed a virtual semi-structured 426 qualitative interview with an unfamiliar research staff member. The service providers had 427 varying backgrounds including two staff with Psychiatric Outreach who referred clients to the 428 study, one manager at a shelter, and one manager at a community day program, both of whom 429 facilitated study activities. 430

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint Study providers identified no difficulties in communicating with the study team. Generally, 432 expectations were met with respect to each individual's anticipated role in the study. One 433 service provider identified that additional guidance on who might be a good candidate and what 434 role they need to play after enrolment in the study would be beneficial. All four service providers 435 felt that both study participation and the delivery of NET integrated well with the services that 436 they currently offer. Clinical providers felt that it was beneficial to have resources that they could 437 offer these clients, in particular noting the benefit to clients not having to attend the hospital for 438 care. 439

The service providers unanimously expressed the need for this service within their community. 441

One provider when asked about the need for NET as a service noted: "It's just more than 442 needed, it's irreplaceable, without it there's nothing. Nothing, like, is really happening right now 443 so it would be great to implement this in a community setting on a broader scale." Similarly, a 444 second provider noted that NET was "absolutely necessary, nothing to really add, there needs 445 to be this type of service for this population, the challenge isn't figuring out if there's a need but 446 more how can we provide what's needed so definitely a need." 447

The service providers were supportive of both the conduct of a large-scale trial all expressed 449 interest in participating in such an initiative. With respect to how to offer NET on a larger scale, 450 the lack of resources, both time and trained staff, was a common concern, however they were 451 all amenable to a centralized resource as an option to delivery therapy services in the 452 community as well. Despite the potential staffing constraint, the service providers indicated their 453 willingness to navigate any barriers that may exist to continuing to offer this therapy in their 454 respective settings. All indicated interest in having a role in further research studies and were 455 keen to facilitate this research. 456

Generalizability 458

The sample enrolled in the study is largely representative of the homeless population, both in 459

Ottawa and nationally. It is reasonable to assume that our results may be generalizable and is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ;  https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint students supervised by experienced therapists is also a model that needs to be explored 478 especially given most universities' commitment to social accountability in their communities. 479

While our initial target of 75% retention was not met, our retention rate is consistent with 480 literature around engagement in psychotherapy, while our proportion of participants for 481 completing all sessions of therapy (80%) exceeded average retention rates(43,44). Given the 482 considerations of working with individuals who are homeless and the potential challenges of 483 engaging with trauma therapy, we feel that these rates are high enough to deem the therapy 484 and participation in a study to be acceptable. 485

What is striking is that the drop-out rate before starting therapy is much higher than after starting 486 treatment, with12/22 (55%) of potential participants dropping out prior to therapy compared to 487 only two out of ten (20%) after starting therapy. It is likely that this is due to a combination of 488 three factors. First is that referrers to the study need to be given clear guidance as to who is is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint Substance use at the time of enrolment did not appear to impact whether or not a participant 503 was likely to complete therapy and follow-up sessions, suggesting that it is feasible to include 504 those using substances in research and therapy. However, staying in a shelter at the time of 505 enrolment did approach significance with respect to not completing the study. Given this, 506 consideration should be given when working with individuals who are staying in a shelter on 507 how to best support therapy retention. Factors to increase retention should be examined during 508 the optimization phase of MOST, with a particular focus on those who are in the shelter or 509 sleeping rough. 510

Within subject differences showed a significant and clinically meaningful reduction in PTSD 511 symptoms in participants from baseline to week 12, suggesting that Narrative Exposure Therapy 512 may be effective in this population. Further research through a large scale RCT is needed to 513 determine the effect of genealogy and optimize the delivery of NET in this population. 514

The major strength of this study is that we have shown that it is both feasible and acceptable to 516 conduct an RCT for the treatment of trauma with individuals experiencing homelessness in a 517 community settingthe first study of its kind. Additionally, the study also shows preliminary 518 evidence for NET as an effective therapy for complex trauma in the homeless. However, the 519 study is not without limitations. The major limitation of this study was a lack of individuals trained 520 in NET. As a pilot feasibility study, the trial is also underpowered to detect any significant 521 differences between treatment groups. 522

With respect to the genealogy report, it took approximately 4-6 weeks to receive a completed 523 report, which made it challenging to incorporate the findings of the report into the narrative 524 process. It would be possible in a large-scale trial to ensure that the individual is seen by the 525 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint genealogist in advance of initiating therapy to increase the likelihood of receiving the report 526 during active narration. 527

Lastly, due to the sudden lockdown measures implemented during the COVID-19 crisis the 528 study team was unable to complete the planned qualitative interviews about participant 529 experiences in the study, the therapy or genealogy support. Anecdotally, there were some 530 complaints about the complexity of two questionnaires (SF-20 and Health Care Costs 531

Questionnaire), but participants did not refuse to answer any questionnaires or comment on the 532 length of visits. With respect to the genealogist, participants all received their reports, with one 533 participant lost to follow-up requesting their report several months later. Only one participant, at 534 the time of consent, indicated that they would have no interest in speaking with a genealogist if 535 they were randomized to that arm. 536

There is currently a significant gap in trauma treatment for the homeless community, despite the 538 high prevalence and degree of complexity within this population. This study has shown that not 539 only is it feasible to deliver community-based therapy without a fixed location, but that Narrative 540

Exposure Therapy is an effective and acceptable therapeutic option for individuals experiencing 541 complex trauma and homelessness. Additionally, service providers have emphatically endorsed 542 the need for this program after their participation in the study. 543

This study was the first of three components of the preparation stage of a multiphase 545 optimization strategy (MOST). The second is a scoping review (46) which we have published 546 separately on the treatment of PTSD in the homeless including the use of trauma informed care 547

to deliver such therapy. The third is a consultation with key stakeholders, including individuals 548 with lived experience, about the implementation of a trauma informed care model for individuals 549 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint who are homeless or vulnerably housed. A focus of this consultation has been addressing the 550 issue of trust when referring potential participants to a treatment study. We will use these 551 components to develop a model of delivering NET in this population which we will optimize 552 using a factorial study in preparation for a larger multi-site RCT to evaluate the efficacy of NET 553 as well as genealogy support in this population. The multi-site RCT will also help to evaluate 554 issues of generalizability across communities that have different supports for their homeless 555 population. 556

NET -Narrative Exposure Therapy 558 

Not applicable. 572

The datasets used and/or analyzed during the current study are available from the 574 corresponding author on reasonable request. 575

The authors declare that they have no competing interests. 577

Funding 578 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 11, 2021. ; https://doi.org/10.1101/2021.11.08.21266074 doi: medRxiv preprint This work was supported by a grant from the Royal Ottawa Hospital's Institute for Mental Health 579

Research (ROH IMHR). The funders had no role in the review or approval of this manuscript for 580 publication. 581

NEE was involved in protocol development, study conduct, data analysis and was a major 583 contributor in writing the manuscript. AB was involved in study conduct. NSD was involved in 584 study design, protocol development, and study conduct. SEM was involved in study design and 585 SH was involved in obtaining funding, study design, protocol development, conducting therapy 586

sessions, data analysis and interpretation, and was a major contributor in writing the 587 manuscript. All authors read and approved the final manuscript. 588

Highlights: 601 Preliminary results from the second nationally coordinated Point-in-Time count of 602 homelessness in Canadian communities

-Kingston-PiT-Count-Results.pdf 608 4. Niagra Region Homelessness Services. Homelessness Point-in-Time Count Report -609 Niagra Region. 2021. 610 5. US Department of Housing and Urban Development. The 2020 Annual Homeless 611

Homelessness and health

Homeless persons' experiences of 614 health-and social care: A systematic integrative review

The health 618 and housing in transition study: A longitudinal study of the health of homeless and 619 vulnerably housed adults in three Canadian cities

Experience of healthcare among the homeless and vulnerably 621 housed a qualitative study: Opportunities for equity-oriented health care

Homeless Patients Associate Clinician Bias With Suboptimal Care 624 for Mental Illness, Addictions, and Chronic Pain

Homeless people's perceptions of welcomeness and 626 unwelcomeness in healthcare encounters

Suicide-related 628 presentations of homeless individuals to an inner-city emergency department. General 629 Hospital Psychiatry

Re-conceptualising approaches to meeting the health needs of 631 homeless people

Lifetime Prevalence of Trauma among Homeless 633 People in Sydney. Aust New Zeal J Psychiatry

Is there Complex Trauma Experience typology for 636

Australian's experiencing extreme social disadvantage and low housing stability?

Trauma and post-traumatic stress disorder among homeless adults 639 in Sydney

The unmet health care needs of 642 homeless adults: A national study

Traumatic Stressor Exposure 646 and Post-Traumatic Symptoms in Homeless Veterans

Homelessness-Related Traumatic Events and 650 PTSD Among Women Experiencing Episodes of Homelessness in Three U.S. Cities. J 651 Trauma Stress

Posttraumatic stress disorder 654 and substance use disorder comorbidity in homeless adults: Prevalence, correlates, and 655 sex differences

Post-traumatic stress disorder in 658

Homelessness as psychological trauma: Broadening 660 perspectives

Shelter from the storm: Trauma-informed care in 662 homelessness services settings

World Health Organization. International Classification of Diseases for Mortalility and 665 Morbidity Statistics Eleventh Edition. World Health Organization

Using latent class analysis to 667 support the ICD-11 complex posttraumatic stress disorder diagnosis in a sample of 668 homeless adults

Traumatic experiences

ICD-11 complex PTSD, and the overlap with ICD-10 diagnoses

Common 673 trust and personal safety issues: A systematic review on the acceptability of health and 674 social interventions for persons with lived experience of homelessness

Universal health 677 insurance and health care access for homeless persons

Narrative exposure therapy : A short-term treatment for 679 traumatic stress disorders. Hogrefe

Clinical Practice Guideline for the Treatment of 681

Crombach A, Elbert T. Controlling Offensive Behavior Using Narrative Exposure Therapy: 685 A Randomized Controlled Trial of Former Street Children

A 687 comparison of narrative exposure therapy and prolonged exposure therapy for PTSD. 688 Clinical Psychology Review

Mixed Methods-Theory and practise.Sequential, 690 explanatory approach

The transgenerational trauma and resilience genogram

A Journey to the Past

Introduction: Indigenous Perspectives on Genealogical Research

Te Ira Tangata: a Zelen randomised 701 controlled trial of a culturally informed treatment compared to treatment as usual in Māori 702 who present to hospital after self-harm

CONSORT 2010 statement: Extension to randomised pilot and feasibility trials. Pilot 705 Feasibility Stud

Reliability and validity of the Addiction 707

Severity Index with a homeless sample

Sample size of 12 per group rule of thumb for a pilot study

EARLY 715 WITHDRAWAL FROM MENTAL HEALTH TREATMENT: IMPLICATIONS FOR 716 PSYCHOTHERAPY PRACTICE. Psychotherapy. 2008. 717 44. Swift JK, Greenberg RP. Premature discontinuation in adult psychotherapy: A meta-718 analysis

Using a trauma-informed, 720 socially just research framework with marginalized populations: Practices and barriers to 721 implementation

Interventions to treat post-723 traumatic stress disorder (PTSD) in vulnerably housed populations and trauma-informed 724 care: A scoping review

The study team would also like to recognize the following individuals for their contributions to 590 the project: Mags Gaulden (genealogist), Kim Bulger (therapist), research staff Brooklyn Ward, 591 and the service users who both participated and advised on the design and conduct of this 592 study. The study team would also like to thank the Royal Ottawa's Psychiatric Outreach 593Program, Ottawa Inner City Health, Centre 454, and the Ottawa Salvation Army (Booth Street) 594for their administrative support of the trial. 595