Measuring PrEP Preferences Among At-Risk Military Populations: Results of an Adaptive Choice Based Conjoint Analysis Study

BACKGROUND . Pre-exposure (PrEP) health matching preferences outcomes. Therefore, an early step for planning program design is to characterize and identify preferences for improved PrEP delivery within this population. 53 METHODS . HIV-negative military MSM were recruited through a closed, LGBT 54 military social media group. Participants completed an anonymous survey presenting five 55 experimentally varied attributes of interest related to PrEP administration (dosing 56 method, provider type, visit location, lab work evaluation location, and dispensing 57 venue). Relative importance and part-worth utility scores were generated using 58 Hierarchical Bayes (HB) estimation, and the randomized first choice model was used to 59 examine participation interest across eight possible PrEP program scenarios. 60 RESULTS . Among the 429 participants and the eight scenarios that varied the five 61 attributes into delivery profiles, the most preferred scenario (69.9%) featured a daily pill 62 or long-term PrEP injection, military provider, smartphone/telehealth visit, lab evaluation 63 on-base, and on-base medication pick-up. Responses indicated the need for providers to 64 familiarize themselves with PrEP prescription knowledge and to provide interactions 65 sensitive to sexual identity, mental health, and decision autonomy. CONCLUSION : These results suggest that a military setting is preferred over a civilian 67 or off-site one, yet more importantly, it points to a high value placed on the quality of 68 clinical interactions. High interest in long-acting implants and injections also suggest 69 potential markets for future PrEP engagement.

Pre-exposure prophylaxis (PrEP) is a medication that effectively prevents HIV infection when taken correctly yet is underutilized within the military health care system. For this study, we identified the preferred characteristics of an ideal PrEP delivery program within the military health care system through an anonymous, conjoint analysis survey of at-risk, U.S. military members. Our findings reveal that a military, on-base health care setting is most preferred for PrEP delivery, yet also indicate a need for quality clinical interactions that are sensitive to sexual identity, mental health, and decision autonomy.
Given the low uptake of PrEP by at-risk military members, we believe that the findings presented in our paper will appeal to key infectious disease stakeholders and policymakers who subscribe to Military Medical Research. Our results will allow your readers to understand which characteristics of a military PrEP delivery program are most attractive to the distinct preferences of this vulnerable population. In doing so, we hope our research will advance evidence-informed PrEP delivery programs within the military health care system tailored to the preferences of U.S. military members most at risk for acquiring HIV.
All listed authors confirm that this manuscript has not been previously published, nor is it currently under consideration by any other journal. Furthermore, all authors have approved the contents of this paper and have agreed to MMR's submission policies.
Should our manuscript be selected for peer review, we would like to suggest the following reviewers that would be able to objectively evaluate our findings and interpretation based on their research background and expertise.

pill for two days after], rectal douche [before sex], injection [every 2 months], implant 140
[once a year]), provider type (military, civilian), visit location (on-base, off-base, 141 smartphone app), dispensing venue (on-base, off-base, mail delivery), and lab 142 evaluation (on-base, off-base, home-based mail kit). The survey was piloted with 11 143 members within the targeted social media group for concept testing, and the descriptions 144 and wording of three attribute categories and two attribute level choices were revised for 145 clarification based on feedback. Additionally, we collected demographical data to include 146 age, race, ethnicity, rank type (officer, enlisted or warrant officer), military branch, 147 geographic region, PHQ2(39) and HIRI-MSM risk score,(40) as well as measures to 148 explore levels of disclosure discomfort and anticipated stigma within interactions with a 149 health care provider. 150

ANALYSIS 151
The final survey instrument was loaded into Lighthouse Studio 9, and an 152 experimental design module was used to pre-test the design with 500 simulated 153 respondents for optimal choice task configuration. The final design produced a survey 154 where each level within an attribute was seen at least three times per respondent; 155 achieving a high degree of precision at the individual level with a standard of error of 156 <0.03 and all efficiencies reporting at 1.00.(41) 157 Table 1 displays the CONSORT diagram of respondent enrollment and exclusion. 158 To ensure the integrity of the data and eliminate random or duplicate responders, security 159 features within the Sawtooth software and servers recognize returning study participants 160 through the use of internet browser cookies and IP addresses. It also prevents repeated or 161 duplicate attempts to retake the survey.(42) Additionally, as extensive pilot testing 162 required at least 10 to 15 minutes, responses completed in less than 10 minutes (or if a 163 respondent selected the same answer for all items) were excluded. Also, the root 164 likelihood (RLH) fit statistic for each respondent was analyzed to evaluate within-165 respondent choice consistency. RLH, which has a probability value from 0 to 1.0, was 166 used to discriminate between respondents who answered choice-questions consistently or 167 randomly.(43) The survey design was tested by 1,000 computer-generated mock 168 respondents to determine the median RLH for 'random responders' at the 95% percentile 169 (0.5178 RLH). Survey respondents with an RLH below this score were excluded, as the 170 inclusion of 'random responders' can affect the calculation of preference scores and 171 participation rates. (43)  172 For conjoint analyses, the Hierarchical Bayes (HB) procedure was used to 173 estimate part-worth utility scores (PWUS) on an individual level for its accuracy and 174 efficiency, (44,45) and was used to analyze the PWUS of the aggregated sample across 175 all 16 attribute levels. The resulting PWUS of the levels under each attribute category are 176 zero-centered; meaning that the sum of the level scores under each attribute category 177 equal to zero. Scores that are further away from zero (0)  influence that each attribute category has on the respondents preference decision-making. 182 The RIS for this study was calculated by dividing the range of PWUS for levels under 183 each attribute by the sum of the ranges, and then multiplying by 100.(46, 47) Therefore, 184 if an attribute RIS is 45%, then this means that 45% of an individual's decision making 185 for product engagement will be influenced by preferences within that attribute category. 186 The PWUS were then used to predict the rate of participation among eight hypothetical 187 PrEP program scenarios. PrEP program scenarios were configured after a variety of 188 currently available or currently feasible PrEP program models, as well as best-and worst-189 case scenarios based on the highest and lowest PWUS among the attribute levels.

RESULTS 201
Participants. Table 2 shows the descriptive statistics of the 429 respondents that 202 met the required elapsed survey time and RLH consistency cut-off. Overall, mean age 203 was 30 years old, 96.7% identified as cis-gendered male (2.6% identified as trans-female, 204 and 0.7% identified as trans-male), 72% were white, 72.5% were of non-Hispanic 205 ethnicity, 46.4% were of officer rank, 54.1% had at least a bachelor's degree or above, 206 and 48.7% were within the U.S. Army branch. Overall, 62.7% screened positive for 207 depressive symptoms, 89.3% were defined as having a high objective risk for acquiring 208 HIV,(40) and 83.0% reported condomless receptive anal sex within the prior six months. 209 In interactions with their primary care provider (PCP), 36.8% were "somewhat-" 210 or "extremely" uncomfortable with talking about sex with their PCP, 48.1% were 211 "somewhat" or "very" fearful of being judged by their PCP for their gay/MSM identity, 212 and 45.2% were fearful for becoming mistreated by their PCP for their gay/MSM identity 213 as well. Furthermore, 64.1% of respondents found it "somewhat" or "very" important that 214 their PCP affirms or show interest in the participant's sexual identity concerning their 215 care, and 78.4% of members found it important that their PCP provides a high degree of 216 medication decision-making autonomy for taking PrEP. 217 Table 3 shows the relative importance scores (RIS) of the five attributes, and 218 Table 4 shows the part-worth utility scores (zero-centered) for each attribute level. For 219 this study, the dosing method was the most critical attribute among the participants with a 220 relative importance score (RIS) of 45.2%; suggesting that the participant's decision-221 making process to participate in a PrEP program is most influenced by the level choice 222 within the dosing method attribute. For dosing method, a daily pill was the most preferred 223 option, although the bi-monthly PrEP injection and yearly implant were also preferred to 224 a slightly lesser degree. The on-demand pill regimen and before-sex rectal PrEP douche 225 were less preferred within the aggregate sample. 226 The provider type attribute was the second most important attribute to 227 respondents, although to a much lesser degree at 15.8% (RIS). Looking at level within 228 this attribute, there was a higher preference for a military than a civilian healthcare 229 provider. The PrEP visit location attribute was the third most important attribute 230 (RIS=14.5%). Respondents preferred to have a virtual medical visit through a smartphone 231 app or on-base location more than an off-base visit location the most. For the laboratory 232 evaluation location attribute (RIS=13.4%), participants preferred to provide specimens 233 for assays to initiate or continue PrEP on-base rather than a location off-base, or through 234 a self-collected, home-based mail-in kit. PrEP dispensing venue had the least influence 235 on participants' decision-making (RIS=11.0%), with participants preferring to receive or 236 pick-up their PrEP medication on-base over a mail delivery service or a location off-base. The significant findings from this study reveal that respondents prefer the 275 convenience of daily pill PrEP services on-base over civilian and off-base settings, yet 276 also indicate a priority to address MSM-specific needs in the context of their care. 277 Despite an overall willingness to disclose same-sex activity, almost half of respondents 278 were fearful of being judged or mistreated by their PCP for their gay/MSM identity. 279 Additionally, over half of the respondents had a positive screening score for depressive 280 symptoms, and the majority of members engage in risk behaviors that categorize them as Results of the conjoint experiment found the dosing method attribute to be the 291 most critical and influential preference factor within a PrEP delivery program. The strong 292 preference for a daily pill (and when available, the PrEP injection and PrEP implant) 293 suggests that a demand remains for alternative short and long-acting PrEP methods 294 within this population. The apparent benefit to long-acting agents is that it lifts the burden 295 of a daily pill from a user; an advantage for an individual with adherence concerns or an 296 This study provides an initial description of the preferences and interest for PrEP 328 by U.S. military service members with a high risk of acquiring HIV. Our results indicate 329 that PrEP interest among this population is most likely to be successful when PrEP is 330 offered as a daily pill, injection, or implant, with a medical visit performed with a 331 military healthcare provider through a telehealth smartphone app. Additionally, allowing 332 on-base locations to provide laboratory samples and to receive PrEP medication can also 333 facilitate program preference. PrEP engagement will further be enhanced by ensuring 334 that medical providers and facilities are knowledgeable and comfortable prescribing PrEP 335 services. Offering an affirming environment sensitive to health care concerns related to 336 mental and sexual health will also be important, as well as provide participants the 337 decision autonomy to take PrEP without pressure. Consequently, key populations, 338 stakeholders, and policymakers will be better equipped for scale-up of PrEP among at-  1: Inclusion criteria demographics were assessed twice; at consent screen for eligibility, and again after conjoint experiment. 2: Omitted responses indicated a service impossibility, such as self-identifying as an Air Force warrant officer (does not exist) 3: PWUS calculated using omitted respondents due to RLH cut-off to examine differences in scores. 4: Increase in NONE utility score (up from -57.7), indicating that including the omitted responses with low RLH scores into the final sample affects the computation of PrEP participation rates calculated using the NONE utility score.    Notes: e: Zero-centered part-worth utility scores imply the positive or negative magnitude of the participant's preference for the level choice in relation to the other level options within the same attribute. f: The "None" parameter represents the positive or negative magnitude in which a respondent is likely to select "None" (not willing to take PrEP in any scenario despite program configuration)