https://orcid.org/0000-0002-3080-366X
Abstract
The Push–Pull–Infrastructure (PPI) model is proposed as a strategy for disseminating health innovations. Using a PPI model, we developed and examined the feasibility of a community-based intervention to disseminate oral human immunodeficiency virus (HIV) self-testing (OHST) to men who have sex with men in Portland, OR.
We disseminated OHST kits through a network of commercial businesses (n = 6) serving the Lesbian, Gay, Bisexual, Transgender, and Queer plus community. Data were collected weekly on the number of kits distributed. Street intercepts were conducted with customers to assess the impact of promotional efforts on intervention awareness. Using a quasi-experimental design, we examined variation in the intensity of the promotional component on OHST dissemination.
Over a 24-week period, we disseminated 2698 OHSTs. Dissemination rates were found to vary significantly with the intensity of the promotional campaign in a dose–response manner (e.g. high intensity = more dissemination). Customer awareness of the campaign increased significantly in the first 9 weeks (from 22% to 60%, P < .001). Within the geographic area studied, we found our dissemination rate to be comparable to or in excess of rates obtained by other county-level HIV-testing programs (i.e. 2698 OHSTs vs. 2561 in-person clinic-based tests; vs. 78 OHSTs disseminated online).
A PPI model based on a community commercial network approach was highly successful. The promotional component successfully increased customers’ awareness of the intervention. Formative work, using a single-case changing-intensity design, provided evidence that warrants more extensive research on strategies for mounting a promotional component that builds on the principle of “more for less.”
Lay Summary
The sale of oral human immunodeficiency virus (HIV) self-test kits through retail outlets (e.g. pharmacies) has been unsuccessful. We examined the feasibility of using Lesbian, Gay, Bisexual, Transgender, and Queer plus commercial businesses to distribute free oral HIV self-test kits to their customers. We successfully distributed 2698 kits in a 6-month period, which exceeds the number of free test kits distributed through Internet services (78 kits) and is comparable to the number of tests performed by county health clinics (2561 tests). Additionally, our promotional campaign was successful in increasing customer awareness from 22% to 60% in a 9-week period. A community-network approach is recommended as a supplemental testing system pending further study of the intervention’s ability to be sustained over longer timeframes.
Practice: Lesbian, Gay, Bisexual, Transgender, and Queer plus (LGBTQ+) community-based commercial networks can be used to successfully disseminate free oral human immunodeficiency virus (HIV) self-test kits at a relatively high rate.
Policy: Policy-makers who wish to supplement existing HIV-testing programs to increase the reach of current HIV-testing efforts should consider the viability of LGBTQ+ community-based commercial networks as a dissemination approach in their community.
Research: Future research is needed to examine the sustainability of a community-based commercial network dissemination approach in terms of network participation, costs, and community buy-in over long periods of time.
Introduction
Effective dissemination is essential to closing the translation gap from the development of an innovation to its full implementation in the real world. Our research team has been interested in the challenges of disseminating innovative healthcare, with a focus on oral human immunodeficiency virus (HIV) self-test (OHST) kits [1–5]. Figure 1 provides a dissemination timeline for OHST kits in the USA. The U.S. Food and Drug Administration (FDA) initially approved OHST kits for administration only by trained clinical personnel in 2004 [6]. However, the FDA did not approve OHST for home use by consumers until 2012 [7]. The hope at that time was that these kits would supplement the existing HIV-testing system [1]. Unfortunately, this has not occurred because of high retail costs [1, 3, 8]. Suboptimal OHST uptake reflects a missed opportunity, as research shows home self-testing to reduce barriers to in-person clinic-based testing such as problems with access, inconvenience, privacy concerns, and fears of social stigma [1, 9–15]. Importantly, HIV self-testing facilitates testing among people who have never been tested or infrequently test [16–22]. This is a significant outcome since individuals who were previously unaware of their HIV status (i.e. had never previously tested for HIV) account for 30%–50% of new HIV diagnoses in the USA [23, 24]. Consequently, it is important to identify strategies for effectively disseminating OHST kits.
A brief history of oral HIV self-testing (OraQuick) dissemination in the USA
We utilized a Push–Pull–Infrastructure (PPI) model, developed by Dearing and Kreuter [25], to construct and evaluate a multilevel community-based network intervention to disseminate OHST. We aimed to examine the intervention in terms of its feasibility and acceptability over a period of 6 months. As described further later, the PPI model (and our intervention) has three major components: Infrastructure, Pull, and Push.
Push–Pull–Infrastructure model: Network infrastructure
The infrastructure elements, such as business networks or community health organizations, provide an environment for delivering innovative practices and products (i.e. distribution system). We developed a community-based infrastructure component, composed of a network of commercial establishments that serve the Lesbian, Gay, Bisexual, Transgender, Queer+ (LGBTQ+) community in Portland, OR (see Methods).
Dearing and Kreuter’s [25] theoretical work has not delved deeply into the challenges of constructing community-based infrastructure. Prior work [26] has found that community-based infrastructure, composed of independent health organizations, may be more successful in working collectively when the network is connected through a dedicated communication system. To our knowledge, community-based network communication systems are seldom in place in a manner that would address day-to-day matters (e.g. troubleshooting logistical problems with how to display or store OHST kits). We recruited a community-based HIV/AIDS organization (CBO) to provide intra-network communication to assist in problem-solving among commercial dissemination sites (see Methods).
Push–Pull–Infrastructure model: Pull component
In consumer-based models, Pull reflects the consumers’ values, beliefs, and behaviors regarding the health innovation and dissemination process [25] (e.g. promotional media channels and product site delivery). In this regard, Dearing and Kreuter [25] acknowledge the importance of also obtaining the perspectives of other stakeholders relevant to the consumer group and the innovation. For instance, we examined perspectives from commercial and HIV/AIDS community services regarding OHST (i.e. personnel from commercial dissemination sites, community-based organizations, and the local county health department [see Methods]). Dearing and Kreuter [25] suggest a number of standard social marketing techniques for gathering information from stakeholders. Consistent with a social marketing strategy, community engagement offers a useful approach to understanding consumer perspectives [27–30]. More broadly, community engagement may also invite stakeholders to share their opinions on the methods used for evaluating the dissemination intervention and to participate in Push-related committees (see later).
Push–Pull–Infrastructure model: Push component
Push promotion stimulates diffusion of program awareness and information about the innovation to the community [25]. Consequently, Push may include planning and promotion committees that work with community representatives (i.e. through community engagement). We used a community engagement approach to identify promotion channels and message content appropriate for the community. Our goal for the Push component was to develop a culturally sensitive, multilevel, information-motivation promotional campaign designed to cue, inform, and motivate uptake of OHST. We evaluated the impact of our promotional campaign on consumer awareness. Further, we used a single-case (quasi-experimental) design (i.e. a changing-intensity design) to examine the impact of promotional intensity on kit dissemination rates over time (see Methods). This represented an initial effort to identify the minimum promotional effort needed to achieve high levels of consumer demand.
Methods
Locale and priority population
Our priority population is defined here as men who have sex with men (MSM), including transgender MSM (MSM is not referring to the perception of one’s sexual orientation, but to a HIV transmission factor; i.e. men who have unprotected receptive/insertive anal intercourse with other men). Our promotional campaign focused primarily on this priority population. However, we recognize that the commercial establishments selected for self-test kit dissemination may serve other populations as well. For instance, the participating retail sex store serves a wide range of individuals with different sexual orientations and behavioral preferences, while the bathhouse is exclusively MSM. The My Test/My Choice (MT/MC) project was conducted in Portland, OR, located within Multnomah County, with an estimated 28 489 MSM residents [31, 32]. The vast majority (84%) of Portland’s HIV+ cases are MSM/MSM-intravenous drug users [33].
Intervention design and procedures
Preliminary research
We examined dissemination of OHST kits through health clinics versus LGBTQ+ business establishments in Portland/Multnomah County (unpublished data, 2017). Self-test kits were disseminated through three LGBTQ+ businesses (bathhouse, book/video stores) and three health clinics serving the LGBTQ+ community. Over a 2-week period, 274 kits were distributed from LGBTQ+ businesses compared to 55 kits through clinics. Thus, the rate of distribution was five times higher through LGBTQ+ businesses. No clinic site reached the level of dissemination achieved by any one of the commercial sites. These results point out the limitations of using healthcare settings (alone or in combination) to disseminate OHST kits. For instance, the volume of clients seeking healthcare during any given week is substantially lower than that of the volume of consumers attending LGBTQ+ commercial businesses.
Infrastructure development and procedures
Our network consisted of six businesses: four LGBTQ+ bars/clubs, one bathhouse, and one adult sex store. Network sites were identified based on a geographic survey of local LGBTQ+ businesses, planning committee recommendations, and a rapid ethnography. The adult sex store was the most heterogeneous in terms of customer characteristics (e.g. sexual orientation, gender), while the bathhouse was the most homogeneous. We conducted a series of meetings with management and other personnel of these six participating businesses to (i) explain study goals, (ii) obtain letters of support, (iii) introduce the research team and individuals serving as a communication hub for the network, (iv) discuss issues of promotion, distribution, kit storage, and kit displays, (v) provide a brief study fact sheet, (vi) discuss staff roles and points of contact, and (vii) obtain preliminary schedules of availability.
Communication hub
Cascade AIDS Project (CAP) served as the communication hub for this study. CAP is a local nonprofit organization that has provided over three decades of HIV/AIDS services to Multnomah County. Each site was provided with a 24/7 cellphone number that would connect them with designated CAP employees who would help if any problems arose (direct calls/text messages). We conducted weekly visits to each site to restock kits and interview site personnel. This provided sites a conduit for sharing information regarding problems and their solutions; this information was communicated to the planning committee.
Community engagement and pull-related procedures
Community engagement [27–30] between the planning committee and community segments (LGBTQ+ customers, the larger LGBTQ+ community, commercial dissemination site personnel, and HIV/AIDS health department/CBO managers) was conducted through focus groups and one-on-one interviews.
LGBTQ± customers
We conducted four focus groups with community members recruited from CBOs and commercial business sites (total n = 14; M = 4.7 participants/group; see Table 1). The majority of participants identified as cisgender men (n = 10), either gay or queer (n = 13), and White/Caucasian (n = 10). Participants ranged in age from 22 to 68 years (M = 36 years, standard deviation [SD] = 17 years). Participants provided ideas and feedback on all aspects of the dissemination process and networks, as well as on the promotional program’s content and the types of communication channels specific to LGBTQ+ persons residing in Portland (see Table 2). They also provided ideas and feedback on how to best communicate information regarding OHST testing protocols and linkage-to-care strategies.
LGBTQ+ customer focus groups: participant characteristics (N = 14)
| Demographic characteristics | Demographic characteristics | N | |
|---|---|---|---|
| Age | Years | Ethnicity: Hispanic/Latinx | 2 |
| Mean (SD) | 36 (17) | Racial identity | |
| Range | 22–68 | American Indian/Alaska Native | 1 |
| N | Asian | 1 | |
| Sex assigned at birth | Black/African American | 2 | |
| Female | 4 | Native Hawaiian/Other Pacific Islander | 0 |
| Male | 10 | White/Caucasian | 9 |
| Gender identity | More than one race | 1 | |
| Female | 0 | Other | 1 |
| Male | 11 | Sexual identity | |
| Transgender male | 2 | Gay | 7 |
| Transgender female | 0 | Bisexual | 1 |
| Other | 1 | Queer | 6 |
| Demographic characteristics | Demographic characteristics | N | |
|---|---|---|---|
| Age | Years | Ethnicity: Hispanic/Latinx | 2 |
| Mean (SD) | 36 (17) | Racial identity | |
| Range | 22–68 | American Indian/Alaska Native | 1 |
| N | Asian | 1 | |
| Sex assigned at birth | Black/African American | 2 | |
| Female | 4 | Native Hawaiian/Other Pacific Islander | 0 |
| Male | 10 | White/Caucasian | 9 |
| Gender identity | More than one race | 1 | |
| Female | 0 | Other | 1 |
| Male | 11 | Sexual identity | |
| Transgender male | 2 | Gay | 7 |
| Transgender female | 0 | Bisexual | 1 |
| Other | 1 | Queer | 6 |
LGBTQ+ customer focus groups: participant characteristics (N = 14)
| Demographic characteristics | Demographic characteristics | N | |
|---|---|---|---|
| Age | Years | Ethnicity: Hispanic/Latinx | 2 |
| Mean (SD) | 36 (17) | Racial identity | |
| Range | 22–68 | American Indian/Alaska Native | 1 |
| N | Asian | 1 | |
| Sex assigned at birth | Black/African American | 2 | |
| Female | 4 | Native Hawaiian/Other Pacific Islander | 0 |
| Male | 10 | White/Caucasian | 9 |
| Gender identity | More than one race | 1 | |
| Female | 0 | Other | 1 |
| Male | 11 | Sexual identity | |
| Transgender male | 2 | Gay | 7 |
| Transgender female | 0 | Bisexual | 1 |
| Other | 1 | Queer | 6 |
| Demographic characteristics | Demographic characteristics | N | |
|---|---|---|---|
| Age | Years | Ethnicity: Hispanic/Latinx | 2 |
| Mean (SD) | 36 (17) | Racial identity | |
| Range | 22–68 | American Indian/Alaska Native | 1 |
| N | Asian | 1 | |
| Sex assigned at birth | Black/African American | 2 | |
| Female | 4 | Native Hawaiian/Other Pacific Islander | 0 |
| Male | 10 | White/Caucasian | 9 |
| Gender identity | More than one race | 1 | |
| Female | 0 | Other | 1 |
| Male | 11 | Sexual identity | |
| Transgender male | 2 | Gay | 7 |
| Transgender female | 0 | Bisexual | 1 |
| Other | 1 | Queer | 6 |
Schedule of promotional components
| Big Push | Little Push | Tiny Push | ||||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Promotional component | ||||||||
| Print materials at distribution sitesa |  | |||||||
| Instagram promotionb |  | |||||||
| Radio PSA (Xray.FM radio)c |  | |||||||
| Social media ads (Sniffies)d |  | |||||||
| Print media (PDX Mercury online newspaper)e |  |  | ||||||
| Street promotion (palm cards)f |  |  | ||||||
| Community event promotiong |  | |||||||
| Big Push | Little Push | Tiny Push | ||||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Promotional component | ||||||||
| Print materials at distribution sitesa | | |||||||
| Instagram promotionb | | |||||||
| Radio PSA (Xray.FM radio)c | | |||||||
| Social media ads (Sniffies)d | | |||||||
| Print media (PDX Mercury online newspaper)e | | | ||||||
| Street promotion (palm cards)f | | | ||||||
| Community event promotiong | | |||||||
aPrint campaign (rotated every three intervals) ran continuously over 12 cycles (see Methods); all materials in English and Spanish.
bIncluded posting each campaign flyer (English and Spanish) on the MT/MC Instagram page at the start of each new print cycle. Posts tagged Cascade AIDS Project (CAP) and five distribution sites (e.g. posts linked to the Instagram page for five sites that had an account); CAP and/or the sites would often re-share our post on their Instagram page. Posts about kit availability at various events/parties during the intervention were re-shared from a site’s page to the MTMC Instagram page, with text describing kit availability.
cRan 10x/week; one PSA recorded in English and Spanish (rotated throughout campaign).
dAd ran from Monday to Wednesday per vendor’s estimation of busiest days (one design [one photo and one message]) in English.
eOn “The Latest” (daily news page) on Monday and Wednesday to Friday; the Savage Love column (long-standing sex and relationship advice column), and the weekly Savage Love email newsletter on Tuesday (one design in English [one photo and one message]).
fPromotional business cards handed out during street intercepts (two designs in English [two photos and two messages]).
gPromotion at two film festivals: the Portland Queer Documentary Film Festival (brief presentation in English and business cards on-site) and the Portland Queer Film Festival (flyer and digital ad in English that ran in a slideshow shown before films).
Schedule of promotional components
| Big Push | Little Push | Tiny Push | ||||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Promotional component | ||||||||
| Print materials at distribution sitesa | | |||||||
| Instagram promotionb | | |||||||
| Radio PSA (Xray.FM radio)c | | |||||||
| Social media ads (Sniffies)d | | |||||||
| Print media (PDX Mercury online newspaper)e | | | ||||||
| Street promotion (palm cards)f | | | ||||||
| Community event promotiong | | |||||||
| Big Push | Little Push | Tiny Push | ||||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Promotional component | ||||||||
| Print materials at distribution sitesa | | |||||||
| Instagram promotionb | | |||||||
| Radio PSA (Xray.FM radio)c | | |||||||
| Social media ads (Sniffies)d | | |||||||
| Print media (PDX Mercury online newspaper)e | | | ||||||
| Street promotion (palm cards)f | | | ||||||
| Community event promotiong | | |||||||
aPrint campaign (rotated every three intervals) ran continuously over 12 cycles (see Methods); all materials in English and Spanish.
bIncluded posting each campaign flyer (English and Spanish) on the MT/MC Instagram page at the start of each new print cycle. Posts tagged Cascade AIDS Project (CAP) and five distribution sites (e.g. posts linked to the Instagram page for five sites that had an account); CAP and/or the sites would often re-share our post on their Instagram page. Posts about kit availability at various events/parties during the intervention were re-shared from a site’s page to the MTMC Instagram page, with text describing kit availability.
cRan 10x/week; one PSA recorded in English and Spanish (rotated throughout campaign).
dAd ran from Monday to Wednesday per vendor’s estimation of busiest days (one design [one photo and one message]) in English.
eOn “The Latest” (daily news page) on Monday and Wednesday to Friday; the Savage Love column (long-standing sex and relationship advice column), and the weekly Savage Love email newsletter on Tuesday (one design in English [one photo and one message]).
fPromotional business cards handed out during street intercepts (two designs in English [two photos and two messages]).
gPromotion at two film festivals: the Portland Queer Documentary Film Festival (brief presentation in English and business cards on-site) and the Portland Queer Film Festival (flyer and digital ad in English that ran in a slideshow shown before films).
HIV/AIDS CBO managers
We conducted one focus group with four participants representing organizations providing HIV-related services in Portland. Two participants were White, and two were Hispanic/Latinx.
Commercial site personnel
Multiple one-on-one interviews were conducted with owners/managers (n = 6) and staff (n = 20). These interviews obtained information and ideas regarding all aspects of the dissemination process and promotional campaign content, as well as information concerning their clientele, customer flow over a typical week, and a variety of practical concerns described above under Infrastructure Development. We utilized personnel input to tailor intervention procedures and materials to each network site (e.g. locations for displaying and storing kits; selection of specific promotional designs and print materials; procedures for restocking kits and maintaining kit displays).
Focus group facilitators/interviewers
All facilitators (n = 3) and interviewers (n = 2) received a 3-day training session. Two focus group facilitators were utilized within each focus group; one focus group was Spanish-speaking only. (Note: Not all Spanish-speaking participants identified as Hispanic/Latinx.)
Push-related procedures
The Push component was composed of a planning committee and a committee that facilitated the development of promotional materials based on feedback from the community engagement process (Pull). The planning committee consisted of members of the core research team and personnel from CAP, the Multnomah County Health Department, and the Oregon Health Authority’s HIV/AIDS Division. Approximately half of this committee were members of the LGBTQ+ community.
Promotional principles and strategies
We promoted the MT/MC Intervention through seven different communication channels identified through our community engagement process as capable of reaching diverse segments of the MSM population (see Table 2). Multilevel promotional efforts are an important consideration because they involve a diverse array of communication channels to provide broad coverage of a given community [27, 34, 35]. As a result, we followed five basic principles in constructing our promotional campaign: (i) use of evidence-based mass media research to guide the work [27, 35], (ii) use of multiple communication channels to reach a diverse audience (see Table 2), (iii) development of culturally sensitive messages and images of relevance to Hispanic/Latinx, Black/African American, Asian, Native American, and White LGBTQ+ communities, (iv) messages incorporating redundancy and novelty to ensure a consistent, fresh message, and (v) messaging in English and Spanish. In addition, the messages provided information on the location of dissemination sites and information regarding OHST that highlighted the relative advantage, simplicity, and trialability of OHST. A majority of our visual and graphic materials were obtained through an extensive search of currently available materials relating to OHST and other HIV-related programs. We obtained information from seven external sources (e.g. Virginia Department of Health, Iowa TelePrEP Program, Connecticut Department of Public Health, and Nebraska AIDS Project). Selected materials were evaluated by our community engagement process.
Promotional campaign development
We developed the promotional campaign by building on recommendations and prior pilot work conducted by members of our planning committee (e.g. Multnomah County Health Department) regarding effective promotional strategies for reaching broad segments of the Portland LGBTQ+ community, and by integrating feedback from community focus groups. In general, the focus group protocols were structured around sets of images and messages that participants reacted to (protocols are available by request from the first author). The structured protocols allowed for the generation of innovative and creative ideas by participants. Focus group data were organized and coded to prioritize and categorize participants’ evaluations of promotional images and messages (e.g. majority and minority opinions regarding characteristics of images and messages; most and least liked images and messages). Analysis summaries for English- and Spanish-language focus groups were generated and discussed among committee members. Overall, focus group findings reflected uniform and consensus opinions (i.e. we found little divergence regarding images and messages that were appealing and meaningful for participants). We utilized these consensus opinions to design the final promotional materials, which included various combinations of three images and two messages.
Linkage-to-care
We provided links to the Portland/Multnomah County Department of Public Health (DPH) and local CBOs offering confirmatory HIV testing, PrEP, and HIV care services. These were provided through printed brochure materials disseminated at each site, in mass media messages, and with the use of QR codes that linked to specific sites (QR codes provided on all posters and print materials).
Dissemination and promotional intervention design
Portions of the promotional campaign paralleled the approximately 24-week dissemination intervention that began on 17 October 2022 and ended on 4 April 2023. The promotional campaign was divided into three phases, with each phase representing a decrease in promotional intensity: Big Push (7 channels, 6 weeks), Little Push (3 channels, 6 weeks), and Tiny Push (1 channel, 12 weeks). This design represents a single-case, quasi-experimental design (i.e. a changing-intensity design) [36]. As applied here, the single case equals a network rather than an individual. This design allows the investigator to examine for outcome variation as a function of changes in stimulus intensity. We extended the low-intensity phase of our promotional effort, Tiny Push, to simulate lower-cost real-world circumstances (i.e. increasing adoptability and sustainability by DPHs). The Tiny Push simulation reduced promotional efforts to only those occurring at the business sites and reduced all contacts with study personnel except those related to resupplying the sites or handling communications. Table 2 shows the three phases of our promotional program and the channels used in each phase.
Evaluation design and procedures
Oral HIV self-test kit counts
Two field staff conducted kit counts one or more times during each interval (≈1 week; except during the final 12 intervals of the study). Kits were labeled with unique identification numbers. We reviewed kits in stock and on display at each site visit, checking this against a list of kits supplied previously (or on hand). This was used to determine the number of kits disseminated. Each site was then supplied with additional kits, such that kits were continuously available at each site (during the final 12 intervals, kits were resupplied; weekly kit counts and interviews with staff were eliminated). During the final 12 intervals (Tiny Push phase), kit counts were obtained at the first and last intervals; using these kit counts and records of the weekly re-supply amounts, we could obtain a total dissemination value for the final 12 intervals. Field staff performed reliability checks on kit counts during weekly/interval visits. Data were uploaded to an electronic data management system.
Program awareness
Customer awareness was assessed using street intercept interviews at two time periods (n = 397). The first round of street intercepts occurred during Weeks/Intervals 1–3 (17 October to 6 November), and the second round occurred during Weeks/Intervals 7–9 (28 November to 18 December). A third round of street intercepts was canceled due to severe weather conditions. Street intercepts targeted days and operation hours with the highest customer volume and were conducted within a block of each business site (with one exception, the bathhouse required intercepts to occur in the business’s foyer). Two to three interviewers obtained brief 2- to 3-minute interviews with each approaching person; interviewers wore identity badges and identifying clothing. Participants were asked if they were going to the business and, if they were, they were offered a $5 gift card for use at the business or a nearby coffee shop if willing to answer two brief questions. First, we asked if they had heard of the My Test/My Choice HIV self-test kit intervention. Second, if they hadn’t heard of it, we asked if they wanted a brief description of the intervention, and, if “yes,” the description was provided. We recorded the number of people who had/hadn’t heard of the intervention. Rather than sampling participants, an attempt was made to recruit all passersby.
Analytic strategies
Data for each site were entered into data management systems (Excel and SPSS) and analyzed with SAS software. Reliability checks were performed at the time of entry by the data management staff. The study director provided additional reliability checks for each time interval and for each commercial site.
Analysis of kit count data
Kit count data were analyzed first by examining overall trends across all intervals. Subsequently, kits were analyzed by each of the three promotional phases; first, using an omnibus test, and then, if significant, by subsequent tests that paired each promotional phase. We also constructed a demand statistic. The demand statistic developed for this project is an adaptation of demand statistics typically found in the field of economics [37, 38] for evaluating product demand. Among other considerations, typical product demand models involve product costs and availability. In the current situation, the product was free and availability was kept relatively constant by continuously resupplying each site. Using data from the current project, a Monte Carlo simulation was conducted using a smooth parametric bootstrap distribution for the median kernel density estimation. The bandwidth of 0.54 was selected under the assumption that the underlying distribution is approximately normal. Specifically, we employed a rule-of-thumb method (e.g. Silverman’s rule-of-thumb) which balances the trade-off between bias and variance in kernel density estimates [39]. This bandwidth is moderate as it is sufficiently narrow to capture the essential features of the density while also providing enough smoothing to mitigate the effects of random fluctuations in our simulated sample (n = 10 000) (This analysis yields a probability distribution that is normally distributed with M = 1.35, Median = 1.33, SD = 0.22, Range = 1.71.) These results support the use of standard inferential statistics. The adjusted demand statistic is defined as the total number of kits taken divided by the product of (i) the number of distribution sites, (ii) the total number of days in each intervention interval, and (iii) the weekly hours of operation, and then multiplied by 100 to produce a whole number. These data were analyzed using a Satterthwaite approximation [40] to estimate the effective degrees of freedom for hypothesis testing (e.g. a two-sample t-test). The demand statistic allowed us to adjust for a number of variable features that occurred in the recording of raw kit counts.
Analysis of program awareness data
Program awareness data were analyzed for each data collection interval and then aggregated again for each data collection period (Intervals 1–3 vs. 7–9). A Z-test for dependent data was computed since customers interviewed during any two interval periods may represent some duplicates.
Results
Intervention awareness
We assumed that the promotional campaign would drive dissemination. Consequently, during the first 12 intervals of the intervention, we evaluated how customer awareness of the intervention was changing over time through street intercepts of customers attending the dissemination sites during Weeks 1–3 (Big Push phase) and 7–9 (Little Push phase). These data are presented in Fig. 2. There was a significant increase in consumer awareness over the first 9 weeks of the intervention from a low of 22% to a high of 60% (Z = 4.7, P < .01; Intervals 1–3 vs. 7–9, n’s, respectively = 253, 144).
Consumer awareness of the MT/MC intervention
Oral HIV self-test kit dissemination
We examined the number of kits disseminated over the course of the three promotional phases (see Table 3). We present kit counts summed over the intervals for both the Big and Little Push phases (first 12 intervals), and present mean values for the Tiny Push phase (last 12 intervals) (i.e. kit counts were computed only for the entire phase, not by interval; see Methods). Comparing the three phases, we found a significant decline in kits disseminated from Big Push to Little Push, to Tiny Push. We distributed 1684 kits over the Big and Little Push phases (56.1% during Big Push, and 43.9% during the Little Push; Z = 7.08, P < .001). As we moved from the Big and Little Push phases to the Tiny Push, the total number of kits disseminated during the 3-month interval covered by the Tiny Push showed a significant decline to 1014 kits disseminated during that period (37.6% of the total kits disseminated; Z = 18.22, P < .001).
Kit dissemination over time × three promotional phases (total kits distributed = 2698)
| Promotional phase | Big Push | Little Push | Tiny Push (mean values)a | |||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Kit distribution count | 511 | 433 | 366 | 374 | 253 | 253 | 253 | 253 |
| Adjusted demand statistic | 4.3 | 4.0 | 3.4 | 3.8 | 2.13 | 2.13 | 2.13 | 2.13 |
| Promotional phase | Big Push | Little Push | Tiny Push (mean values)a | |||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Kit distribution count | 511 | 433 | 366 | 374 | 253 | 253 | 253 | 253 |
| Adjusted demand statistic | 4.3 | 4.0 | 3.4 | 3.8 | 2.13 | 2.13 | 2.13 | 2.13 |
Notes: Adjusted kit demand = Total number of kits taken/([number of distribution points] × [Total number of days in intervention interval] × [Weekly hours of operation]); adjusted demand statistic × 100 to obtain whole numbers.
aMean values are reported for each three-interval period during the Tiny Push phase (see Methods and Evaluation design and procedures); mean values were calculated using the total number of kits disseminated during the Tiny Push phase (n =1014) and dividing by four.
Kit dissemination over time × three promotional phases (total kits distributed = 2698)
| Promotional phase | Big Push | Little Push | Tiny Push (mean values)a | |||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Kit distribution count | 511 | 433 | 366 | 374 | 253 | 253 | 253 | 253 |
| Adjusted demand statistic | 4.3 | 4.0 | 3.4 | 3.8 | 2.13 | 2.13 | 2.13 | 2.13 |
| Promotional phase | Big Push | Little Push | Tiny Push (mean values)a | |||||
|---|---|---|---|---|---|---|---|---|
| Intervals: (≈weekly) | 1–3 | 4–6 | 7–9 | 10–12 | 13–15 | 16–18 | 19–21 | 22–24 |
| Kit distribution count | 511 | 433 | 366 | 374 | 253 | 253 | 253 | 253 |
| Adjusted demand statistic | 4.3 | 4.0 | 3.4 | 3.8 | 2.13 | 2.13 | 2.13 | 2.13 |
Notes: Adjusted kit demand = Total number of kits taken/([number of distribution points] × [Total number of days in intervention interval] × [Weekly hours of operation]); adjusted demand statistic × 100 to obtain whole numbers.
aMean values are reported for each three-interval period during the Tiny Push phase (see Methods and Evaluation design and procedures); mean values were calculated using the total number of kits disseminated during the Tiny Push phase (n =1014) and dividing by four.
The raw kit dissemination data may contain biases because the data collection intervals are not equal over time and hours of operation differed across network sites (Range = 64–138 hours, M = 86.66 hours). The Demand statistic presented in Table 3 controls for these differences. Across all three phases of our promotional effort, Demand is positive throughout but declining across each phase. We compared the mean Demand statistics across each promotional phase using a Satterthwaite approximation to estimate effective degrees of freedom for hypothesis testing using the t-distribution, where mean Demand values equal 4.1, 3.4, and 2.13, respectively (post-hoc tests: Big vs. Little Push, t = 4.12, P < .001; Big and Little Push vs. Tiny Push, t = 14.0, P < .001).
Discussion
Our study used Dearing and Kreuter’s PPI Model [23] to develop and assess a community-based intervention to disseminate OHST among MSM in Portland. The dissemination network (Infrastructure) and promotional campaign (Push) were designed based on input from local public health and LGBTQ+ community members (Pull). We examined the intervention’s impact over a 6-month period in terms of OHST dissemination rates and customers’ awareness of the intervention over time. Overall, the results of the current investigation suggest that a PPI approach using LGBTQ+ commercial settings is a feasible and acceptable method of disseminating OHST.
Community-based networks: dissemination system
Dearing and Kreuter [25] underscore the importance of developing infrastructure to disseminate health innovations. Our study supports this perspective. We developed a community-based network consisting of commercial businesses that serve predominantly LGBTQ+ customers. This dissemination effort was highly successful. To test the relative success of our program, we compared our dissemination data to data from two HIV-testing programs operated by the Portland/Multnomah County DPH (for the October 2022–March 2023 window in which our intervention operated). The total number of kits disseminated by our intervention was 2698 compared to 2561 tests conducted by the county’s in-person clinic-based system, and 78 OHST kits disseminated by the county’s online program for the same period [41]. Consequently, our program disseminated approximately the same number of tests as in-person tests conducted, and substantially more tests than the online OHST program distributed.
In addition to successfully disseminating a large volume of OHST kits in a relatively brief time, the commercial sites also played an important role in tailoring the promotional campaign to their consumer base. Commercial sites stimulated kit demand using attractive well-located displays and through direct promotion by personnel. This level of participation is not new to LGBTQ+ commercial establishments. Since the onset of the AIDS epidemic, LGBTQ+ businesses have played important roles in the dissemination of prevention information and products [42, 43]. These commercial networks have been found to offer LGBTQ+-friendly environments that reduce fears of social stigma and intersect with diverse segments of potentially at-risk persons [43–46]. In this regard, the history of LGBTQ+ commercial businesses makes this type of network intervention an extremely good fit between the infrastructure, the community, and the health innovation to be disseminated.
Push–Pull: developing a multilevel promotional program
We varied the intensity of our promotional efforts using a single-case, quasi-experimental design (i.e. a changing-intensity design) to examine the impact of promotional intensity on kit dissemination rates. We began with a high-intensity program at the start of the intervention (Big Push) that utilized seven different communication channels. This high-intensity effort was successful in increasing diffusion of intervention awareness among network customers. After 6 weeks, this high-intensity campaign was reduced to three communication channels for about 6 weeks, and to one channel for approximately 12 weeks. Kit dissemination was found to show a dose–response curve that declined significantly as promotional intensity decreased. These results, utilizing a changing-intensity design, provide preliminary evidence that warrants further investigation with more sophisticated (and costly) designs. The current design does not control for saturation (e.g. novelty effects that wear off over time) or habituation effects (e.g. declining impact of the promotional program over time), but it is difficult to construe how habituation or saturation would follow the relatively exact dissemination timeline observed here. We also did not control for potential seasonal effects that may occur over a longer time span (e.g. 12 months) or confounding among promotional levels. There is a need for a longer-term study that extends beyond these seasonal windows and varies the intensity of the promotional effort along the lines of an ABAB single-case design [36] (where A = High-Intensity Promotion, B = Lower-Intensity Promotion), repeated over seasonal windows.
Over the last 12 weeks of the intervention, we simulated conditions that might prevail in real-world settings where programs are more likely to be adopted and sustained if costs are low and personnel efforts are minimal [47, 48]. We reduced the number of contacts between staff and work sites, and minimized the promotional program to a low-cost, low-intensity design (see Methods). As noted previously, there was a significant decline over time in kit dissemination, such that in the low-cost/intensity period only 85 OHST kits were disseminated per week, compared to 157 OHST kits/week during the first 6 weeks. These results would suggest that a more moderate reduction in cost/intensity or occasional introduction of high-intensity promotional efforts may be necessary to sustain higher kit dissemination levels. Although the simulation is suggestive, a longer-term, larger-scale test of the minimal-effort promotional campaign is warranted.
Theory: the importance of network communication and community engagement
Although Dearing and Kreuter [25] do not mention community engagement specifically, they clearly understand the importance of identifying characteristics of the target population that will support a successful dissemination effort. In public health, there’s been a long-standing emphasis on the process of community engagement. Community engagement processes recognize that priority populations do not necessarily share the same beliefs and behaviors of the program investigators. Public health programs are routinely tailored to achieve a high degree of fit with important social–psychological characteristics of the target population and other relevant stakeholders. Low-cost methodologies used for social engagement include the use of rapid ethnographic studies, focus groups, one-on-one interviews, and community representation on advisory committees. The challenge of these methodologies is that they’re based on small numbers of observations that may not be generalizable. Consequently, community engagement strategies may not always provide the best answers. Ideally, a probability-based survey of the community would be required to generate a representative understanding of the priority population’s beliefs and behaviors regarding health innovation. These types of complex surveys, however, are expensive and, therefore, not likely to be employed in real-world settings.
Dearing and Kreuter’s [25] three-component dissemination model (Push, Pull, and Infrastructure) may be applied to a number of different contexts. This flexibility in the model may require additional components. In our case, we believe that interorganizational network communication is a necessary component of the model because commercial businesses were not typically in regular contact. Our communication component also served to provide more immediate feedback from the dissemination sites to the planning and promotion committees as to how we might tailor promotional efforts. There is, of course, an additional cost to adding a communication component because personnel are needed to conduct contacts and provide feedback.
Future research directions
Our study provides preliminary evidence that a community-based network intervention offers a feasible and acceptable strategy for disseminating OHST within LGBTQ+ communities. Further research is needed to address the long-term sustainability of a network dissemination system (e.g. ability to sustain network infrastructure), to evaluate the PPI intervention’s ability to reach high-priority populations (e.g. high-risk, never-tested individuals), and to examine the antecedents of OHST uptake in LGBTQ+ commercial environments. Research on program efficacy and maintenance over longer time periods informs the decision to scale up the intervention to other cities with LGBTQ+ commercial networks.
Limitations and strengths
The single-case changing-intensity design used for examining the effect of changing promotional intensity on dissemination is appropriate as an early translation research design. This approach substantially reduced study costs. Alternative case–control studies, such as including control cities that use different promotional intensities, are very expensive and challenging; for instance, it is difficult to match cities in a precise manner. As mentioned previously, there are more complex single-case designs (i.e. a case in this instance refers to the Network) that may be employed in future studies that address limitations in the changing-intensity design. Although the results of the present study are encouraging, the generalizability of the findings to other urban, LGBTQ+ communities is unknown.
We examined changes in customer awareness using time-space sampling (i.e. street intercepts), that were based on work conducted at times of high customer flow. Therefore, the results may not be generalizable to customers who more often frequent commercial sites during low-volume periods. Moreover, longitudinal cohort studies are needed to examine the broader diffusion of promotional efforts among customers, as well as examine the diffusion of kits among friends of customers.
Strengths of this study include efforts to maintain high availability of the kits over the course of the dissemination period to reduce error in the assessment of dissemination rates that might be due to periods in which no kits are available. Additionally, community engagement was successfully employed in this study and led to a very robust acceptance of the program by both customers and personnel at the dissemination sites. Lastly, we would note that the current study utilized an early translational research design that explored key components of the dissemination program and provide insights that inform future investigations.
Conclusions
The current study was highly successful in developing a PPI dissemination intervention built on community-based commercial networks. Preliminary work points the way to conducting more complex investigations to examine alternative strategies for mounting promotional campaigns that build on the general principle of “more for less.” There is also a significant need for research to further examine the efficacy of disseminating large numbers of OHST kits in the LGBTQ+ community in terms of linkage-to-care outcomes. Research that utilizes community engagement strategies and mixed methods techniques is particularly valuable for understanding how consumers and implementers respond to a community-network intervention. Lastly, we believe it’s highly important to understand the extent to which our promotional campaigns and network dissemination efforts are diffused throughout the larger LGBTQ+ community.
Acknowledgements
Special thanks to Kim Toevs, MPH, Multnomah County Health Department retired; Matthew Kreuter, Professor, University of Washington St. Louis; Nancy Vargas, Instructor, Western University; and Jeff Henne MA, CEO of The Henne Group for their significant contributions to the early development of this study.
Funding
This study was funded by the National Institutes of Health (MH120512; PI: M.M. Dolcini). The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Conflict of interest statement
All authors declare that they have no conflicts of interest.
Human Rights
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. All study methods and materials were approved by the Institutional Review Board at Oregon State University.
Informed Consent
Informed consent was obtained from all individual participants included in the study.
Welfare of Animals
This article does not contain any studies with animals performed by any of the authors.
Transparency Statement
Study Registration. The study was not formally registered. Analytic Plan Pre-Registration. The analysis plan was not formally pre-registered.
Analytic Code Availability. Analytic codes used to conduct analyses presented in this study are not available in a public archive. They may be available by emailing the corresponding author.
Materials Availability. Materials used to conduct the study are not publicly available.
Data availability
De-identified data from this study are not available in a public archive. De-identified data from this study will be made available (as allowable according to institutional IRB standards) by emailing the corresponding author.
