A Descriptive Cross-Sectional Analysis Of Potential Factors, Motivations, And Barriers Inuencing Research Participation And Retention Among People Who Use Drugs In The Rural United States

Background: Despite high morbidity and mortality among people who use drugs (PWUD) in rural America, most research is conducted within urban areas. We describe inuencing factors, motivations, and barriers to research participation and retention among rural PWUD. Methods: We recruited 255 eligible participants from community outreach and community-based, epidemiologic research cohorts from April-July 2019 to participate in a descriptive cross-sectional study. Eligible participants reported opioid or injection drug use to get high within 30 days and resided in high-needs rural counties in Oregon, Kentucky, and Ohio. We aggregated response rankings to identify salient inuences, motivations, and barriers. We estimated prevalence ratios to assess for gender, preferred drug use, and geographic differences using log-binomial models. Results: Most participants were male (55%) and recently injected methamphetamine (61%) and/or heroin (57%). The primary inuential factors for research participation are condentiality, amount of nancial compensation, and time required. Primary motivations for participation include nancial compensation, free HIV/HCV testing, and linkage with resources. Changed or false participant contact information and transportation are principal barriers to retention. Respondents who prefer methamphetamines over heroin are more inuenced by why their information is collected and how it is used (PR=1.12; 95%CI:1.00, 1.26). Knowing and wanting to help the research team are motivations for participation among Oregon participants (PR=2.12; 95%CI:1.51, 2.99) and females (PR=1.57; 95%CI:1.09, 2.26). Conclusions: Beyond nancial compensation, researchers should emphasize condentiality, offer testing and linkage with care, use several contact methods, aid transportation, and accommodate demographic differences to improve research participation and retention among rural PWUD.


Introduction
The rural United States (U.S.) is in the midst of an ongoing substance use disorder epidemic. In 2019, one in ve Americans used an illicit drug and 70,630 drug overdose deaths occurred in the U.S, of which 71% were opioid-related [1,2]. Further exacerbating the substance use epidemic is the prediction of the opioid epidemic's '4th wave', or the rise in methamphetamine use among people with opioid use disorders [3][4][5][6]. effective treatments and medical care, and supported linkage to community SUD resources and programs [18,19]. Although clinical trials may provide valuable care and treatment to rural PWUD, retention and recruitment remains challenging [15,20,21]. Those in rural communities note mistrust and fear as contributors to low recruitment, and rural residents have a lower likelihood of awareness of opportunities to participate in research compared to their urban counterparts [21,22].
Factors affecting clinical trial adherence can be categorized using the Ickovics and Mieslers multifactorial framework: the individual, treatment regimen, patient-provider relationship, clinical setting, and the disease [23]. Individual level differences between genders in research recruitment and participation among rural PWUD have been mixed. The National Institute on Drug Abuse (NIDA) Treatment Clinical Trials Network (CTN) did not identify gender differences in research recruitment and retention among PWUD, and willingness to participate in an HCV vaccine trial among PWUD did not differ by gender [24,25]. However, women are still underrepresented in HCV and HIV clinical research [26], and gender differences in utilization of harm reduction services exist among rural PWUD [27]. An individual's drug preference for methamphetamine over heroin may also impact trust in research due to adverse effects such as elevated paranoia and suspiciousness associated with methamphetamine use [28].
Patient-provider relationship factors such as stigmatizing attitudes unique to certain drugs [29][30][31] could result in variations in the perceived judgement by staff as a factor affecting participation.
Our study's primary objective is to describe the in uencing factors, motivations, and barriers of rural PWUD in participation and retention in research studies. The study's secondary objective is to examine variation in in uencing factors, motivations, and barriers across geographic regions, gender, and substance use to inform retention and recruitment strategies.

Study Setting
We recruited participants in rural areas of Kentucky, Ohio, and Oregon, where each state has an established research infrastructure through the National Rural Opioid Initiative [32]. Rural study sites in Eastern Kentucky and Southeastern Ohio are located in Appalachia, a cultural and geographical region that spans 13 states from New York to Mississippi [33]. The Oregon rural study sites in the Paci c Northwest include both coastal and interior communities in large, sparsely populated counties. The populations in these areas of Appalachia and Oregon are predominantly White, and on average 14-25% of the population lives below the poverty line [34]. These rural communities are at increased vulnerability for HIV and HCV transmission due to high injection drug use rates and inadequate healthcare infrastructures [35][36][37].

Study Design
We conducted a cross-sectional, multi-state survey from April-July 2019. The survey was part of the formative phase of the Peer-based Retention of People who Use Drugs in Rural Research (PROUD-R 2 ) study that tests rural peers' ability to improve study retention (ClinicalTrials.gov identi er: NCT03885024) [38]. We derived a sample size of 225 participants to complete the cross-sectional survey and inform the central phase of PROUD-R 2 .
Eligible participants were at least 18 years of age, injected any drug or used opioids to get high within the past 30 days, and lived within rural counties associated with each study site. We recruited participants using convenience sampling at syringe service programs, local health departments, community-based settings, and through concurrent epidemiologic studies. We obtained informed consent from all participants and provided each participant with $20 cash or a gift card as reimbursement for survey participation. The survey was interviewer-administered in Kentucky and Oregon and self-administered in Ohio. In Kentucky and Ohio, we collected data using Qualtrics software, Version June 2019 (Qualtrics, Provo, UT). In Oregon, study data were collected using REDCap electronic data capture tools hosted at Oregon Health & Science University [Grant#: UL1TR002369] [39].

Data Collection
We adapted survey questions from a study that assessed willingness to participate in an HCV vaccine clinical trial among rural PWUD in Appalachia [24]. The Community Advisory Board of the Kentucky CARE2HOPE study and peer recovery support specialists of the Oregon OR-HOPE study reviewed and approved the survey's nal version to con rm the appropriateness of the survey.
Participants provided the following demographic information: age, gender, education, race, and ethnicity. We assessed participant's' gender using the construct, "What is your gender?" in alignment with recommendations to use gender as opposed to sex when reporting psychosocial or cultural factors [40]. We also asked participants about recent drug use, "Have you ever injected drugs to get high?," "Which drugs have you injected in the past 30 days to get high?," and "Which is your drug of choice for getting high?." We collected data on (1) in uencing factors for research participation, (2) motivations for participating in research, and (3) barriers to attend follow-up research appointments through a series of nominal response option questions in which the participant could "select all that apply." To elicit factors that in uenced participants' decision to participate in a research study, we asked, "What are some of the things that people who use drugs in this community may consider when deciding to participate in a research study?" [15 response options]. We inquired, "What are some of the reasons that people who use drugs in this community may decide to participate in a research study?" [10 response options] to obtain their motivations for participation in research. To assess barriers to retention in attending follow-up research appointments, we asked, "What do you think are some of the challenges to getting people to come back for follow-up appointments?" [11 response options]. Complete response options for all three questions are listed in Tables 3-5. If a participant selected more than three responses to the above questions, a second question prompted the participant to rank their rst, second, and third options from their previously selected responses, hereafter referred to as primary, secondary, and tertiary responses.

Statistical Analyses
Participant sociodemographics and drug use characteristics were summarized using descriptive statistics. We excluded participants from the analysis if they did not provide a response for any eligibility criterion. To visually represent and compare the ranking of participant responses, we used diverging stacked bar charts [41]. We displayed survey items with the highest to lowest frequency of primary, secondary, and tertiary responses. We also included counts of selected, but unranked, responses for each survey item.
We aggregated the survey responses into 36 binary dependent variables to analyze the differences of in uencing factors, motivations for participation, and barriers to retention in research among subgroups of the study population. Due to small cell sizes, we condensed the rankings of each response into dichotomous variables ('selected', 'not selected') that represented a participant's response to selecting items as essential in uencing factors, motivations, or barriers.
We assessed differences in the dependent variable by the independent variables of gender (male, female), region (Appalachia, Oregon), and preferred drug of choice (heroin, methamphetamine). We combined Kentucky and Ohio into a single 'Appalachia' group for several conceptual and statistical reasons: the strati ed prevalence ratios of Kentucky and Ohio were similar when compared to Oregon, the Kentucky and Ohio research sites are geographically close, and both are in the Appalachian region. To ensure that "preferred drug of choice" re ected actual use, we veri ed that most participants had access to their preferred drug of choice by generating cross tabulations with their preferred drug of choice and reported substance use in the past 30 days.
We performed log-binomial regression to assess differences in site, gender, and preferred drug use for each of the 36 selected responses. Prevalence ratios (PR) and corresponding 95% con dence intervals (CI) were estimated for each bivariable model. To reduce bias and improve model precision, only survey items with at least ten responses at each level of the binary dependent variable were modeled [42][43][44].
We aggregated and analyzed data using SAS software version 9.4 (SAS Institute Inc., Cary, NC) at Oregon Health and Science University and the Ohio State University. Plots were developed with the "HH" package in R version 4.0.2 (R Core Team, Vienna, Austria) [45,46]. The Ohio State University Institutional Review Board, University of Kentucky Institutional Review Board, and Oregon Health and Science University Institutional Review Board approved this study.

Participant Characteristics
A total of 290 participants completed the survey. In Oregon and Kentucky, a total of 218 participants were screened, and 34 participants were ineligible due to not meeting drug use eligibility criteria (n = 28), living outside of the study location (n = 4), or missing information for all eligibility criteria (n = 2). In Ohio, participant eligibility was assessed verbally, and the number of ineligible participants and reasons for ineligibility were not obtained. We excluded one Ohio participant from analyses due to missing age. The nal analytic sample contained 255 participants and included a complete set of responses for all independent and dependent variables.
The characteristics of the 255 participants included in our study are shown in Table 1. Most participants were from Kentucky (n = 105), then Oregon (n = 79), followed by Ohio (n = 71) and the mean age was 38 years (SD = 9.7). Most participants identi ed as male (55%), white (88%), and had at least a high school diploma/General Educational Development (GED) (73%). Nearly all participants had injected some type of drug in the past 30 days to get high (93%) ( Table 2). Most participants preferred either heroin or methamphetamine; both groups reported recent use (92% for heroin, 95% for methamphetamine), and both were the most commonly injected drugs in the past 30 days (57% and 61%, respectively). Other recently injected drugs included fentanyl (18%), buprenorphine (18%), painkillers (15%), cocaine/crack (8%), methadone (3%), and prescription anxiety drugs (3%).  The primary in uencing factor for research participation was the amount of nancial compensation received in exchange for participation ( Fig. 1), followed by con dentiality of information. Other essential in uencing factors among rural PWUD were the time required to participate in the research study and privacy of the research o ce. The in uencing factor with the least number of ranked responses was whether the research institution or university was well-respected.
In considering participation in research, Oregon respondents had a higher prevalence of selecting all in uencing factors listed compared to Appalachian respondents. When compared to Appalachian respondents, Oregon respondents were more in uenced by how much time is required for participation  How much money they will receive c --How much the project will bene t them overall c --Whether their appointment times will interfere with their work schedule 1.63* (1.24, 2.14) Whether they have childcare so that they can attend their appointments 1.

Patterns of Motivations for Research Participation
Financial compensation was the primary motivator for participation, followed by free diagnostic testing for infectious diseases such as HCV and HIV, linkage to resources, and follow-up testing (Fig. 2). Knowing a person on the research team was the least selected motivator for participation.  (Table 3).
Female participants were nearly twice as likely to report being motivated to participate in a research study if they knew someone on the research team (PR = 1.81; 95% CI:1.09, 2.26), and were marginally more motivated to participate if a nancial incentive was offered (PR = 1.10; 95% CI:0.99, 1.23) ( Table 4).
Respondents whose preferred drug of choice was methamphetamine had a higher prevalence of being motivated to participate in research if they would receive free diagnostic testing (PR = 1.08; 95% CI:0.99, 1.18) and linkage to resources and follow-up testing as part of the study (PR = 1.09; 95% CI:0.99, 1.21), compared to respondents whose preferred drug of choice was heroin ( Table 5).

Patterns of Anticipated Barriers for Retention in Follow-up Research Appointments
Losing contact with participants due to changed contact information had the highest frequency of primary responses among barriers to returning to follow-up research appointments, followed by trouble obtaining transportation and sharing false contact information at their initial appointment (Fig. 3). The barrier with the lowest number of ranked responses among respondents was that they may have stopped using drugs and no longer believe that the study is relevant to them.
Reporting con icts in returning to follow-up appointments due to work, nding childcare, and transportation were greatest among Oregon participants compared to Appalachian participants (PR = Oregon participants also had a higher prevalence of reporting privacy and con dentiality concerns (PR = 1.52; 95% CI:1.22, 1.90) and becoming unreachable due to participation in a drug treatment program (PR = 1.33; 95% CI:1.09, 1.63). Barriers that did not differ between Appalachian and Oregon participants included not being able to get in touch with participants because they provided false contact information and they may be afraid that the staff would judge them if they are still using drugs (Table 3).
Participant-prioritized barriers did not differ by drug of choice (Table 5).

Discussion
We identi ed several themes of rural PWUD considerations in deciding to participate and remain in research studies. The primary in uencing factor and motivator for rural PWUD to participate in research is the amount and presence of nancial compensation. Economic and social factors of the risk environment framework are determinants of substance use [47] and promote a disproportionate burden of substance use among people living below the federal poverty threshold [1]. A lack of assistance programs in areas where rural PWUD reside further exacerbate the economic needs of this population [48,49]. The weight of nancial compensation in our ndings is consistent with studies of rural Kentucky PWUD populations and others that note receipt of nancial compensation as positively associated with research participation and retention [24,50,51]. Lower economic status is also associated with poorer study retention; once enrolled in a longitudinal research study, PWUD who live below the federal poverty line are more likely to be lost to follow-up [52].
Financial need is an undercurrent relevant to other highly noted factors. Transportation was a major perceived barrier to retention among all participants irrespective of participant region, gender, or preferred drug use. Our ndings align with previous studies that note transportation, or distance, as a primary barrier to retention among rural community members [20]. Western Oregon's remote setting may increase transportation challenges. Oregon participants were more likely to consider the frequency of visits and if their friend, family, or partner participates, which can both in uence transportation concerns. Strategies to alleviate transportation challenges might include travel reimbursement, nancial incentive amounts that account for transportation cost, or use of mobile or outreach models that bring the research to the participant.
Although nancial incentive is the primary motivation for research participation among rural PWUD, our study supports other ndings that motivations are multi-dimensional beyond monetary gains such as believing in the mission of the research and seeking linkage to care and other resources [50]. A lack of income paired with scarce medical care in rural locations may also explain the motivations noted by most rural PWUD to participate if linkage with resources and free testing are offered as part of the study.
Due to a lack of healthcare assistance programs, rural PWUD may utilize access to the minimal healthcare offered by clinical trials [48].
Privacy, con dentiality, and interaction quality with research staff are crucial in uencing factors for PWUD in deciding to participate in research, likely due to stigma and the legal, employment, and interpersonal relationship consequences of substance use. Oregon respondents were more likely to be motivated to participate in research if they knew someone on the research staff. This nding may be related to the design of Oregon's concurrent study (i.e., the Oregon HIV/HCV and Opioid Prevention and Engagement, or OR-HOPE) which employs peer recovery support specialists as study staff members.
While we found that most factors among rural PWUD did not differ between males and females, aligning with the ndings of an analysis of 24 NIDA CTN trials [53], we found a notable difference in the importance of privacy between genders. Females reported an increased likelihood of indicating if their information would be kept con dential and whether they can skip questions that make them uncomfortable as important for participation. Female PWUD participants might be more concerned with privacy due to concerns of losing custody of their children if their drug use became publicly known [48] or due to anticipated distress around certain topics related to past trauma. Studies that recruit primarily rural, female PWUD populations should highlight the ability to skip questions and con dentiality protections when obtaining informed consent, and in surveys, questionnaires, and other data collection items to encourage participation.
Changing of participants' contact information was a primary perceived barrier to returning to follow-up appointments, and did not differ between participant region, gender, or preferred drug use. Our ndings align with previous studies that note successfully contacting participants as a barrier to retention among rural community members and PWUD [54]. Obtaining information from participants about contact information of others (family, friends, etc.) who know how to reach them in case they cannot be contacted may improve retention [20]. The challenge of losing contact with participants may be alleviated by providing phone cards or other forms of contact reimbursement.
Participants who reported methamphetamine as their drug of choice to get high, as compared to those who selected heroin, had a higher prevalence of considering factors central to privacy and con dentiality.
While not all of these factors met the threshold for statistical signi cance, the positive measure of association speaks to a theme of distrust and privacy concerns present among those who prefer methamphetamine use. These ndings align with a community-based study in Vancouver, Canada, that found those who use methamphetamine reported greater suspiciousness and paranoia compared to those who use opioids [28]. Methamphetamine-associated paranoia may magnify the general distrust of healthcare systems where PWUD frequently experience stigma [55,56] and may exacerbate skepticism about the transparency of research which is already elevated among rural residents [57]. Research enrolling people who use methamphetamines in rural communities should tailor recruitment and retention strategies to emphasize con dentiality and privacy.
Our ndings should be interpreted in light of several potential limitations. First, though our sample was drawn from U.S. rural communities in three states, ndings may not be generalizable to rural communities outside of Appalachia and southwestern Oregon and may not be representative of all PWUD in the study communities due to our use of convenience sampling for data collection. Second, the numerous response options provide crucial descriptive information on improving clinical trial recruitment and retention in rural areas but are likely correlated. Future work with large population-based samples will be needed for testing multiple hypotheses of multilevel factors. Still, our study found differences in factors for participation and retention between geographic locations and types of preferred drug use. The cross-sectional design of our study is a limitation in regard to capturing the challenges of enrollment and retention over time. We recommend that future longitudinal clinical research studies explore enrolled participants' in uencing factors and motivations for participation. Study staff should collect data on the reasons rural PWUD participants miss follow-up appointments among participants who are not lost to follow-up. Finally, participants were recruited in locations where epidemiological studies had already been recruiting rural PWUD; nearly two-thirds of participants (64% overall, ranging from 42% in Ohio to 75% in Kentucky) reported previously participating in a research study. Therefore, our results may not adequately capture the perspectives of rural PWUD less familiar with or interested in research, which may differ.
However, because our participants are more familiar with research, the reported factors may have been less hypothetical than with research naïve or adverse PWUD.

Conclusions
Our ndings contribute to the CTN's focus on reaching underserved populations, such as rural PWUD, by identifying services such as testing, linkage to care, transportation, and factors such as privacy of clinic location and con dentiality of participant information that may enhance research participation and retention among this population. Research staff can address barriers to returning to follow-up appointments for rural PWUD by providing nancial compensation, collecting detailed contact information from participants, and providing resources for transportation or by bringing the research to the participants through mobile or street outreach. Future longitudinal clinical research can leverage prominent in uencing factors, motivations, and barriers to enhance participation and retention among rural PWUD. and Oregon Health and Science University Institutional Review Board reviewed and approved the study protocol before research activities were initiated. All participants provided informed consent prior to enrollment.

Consent for Publication
Not applicable.
Availability of Data and Materials  Ranked motivators for joining a research study among PWUD in rural communities, April 2019-July 2019