Three FGDs were conducted; two in Atlanta, GA in January 2010 (N = 5 and N = 9) and one in Minneapolis, MN in May 2010 (N = 10). Atlanta and Minneapolis were selected for their divergent geographic locations and HIV epidemics in order to obtain a variety of opinions and healthcare experiences. Eligible FGD participants were male at birth, ≥ 18 years of age, self-reported testing HIV negative, having had sex with a man in the past 12 months, being a resident in the Atlanta or Minneapolis metropolitan areas, and having had at least one visit to a healthcare provider in the past 12 months. Participants completed the FGDs anonymously, thus limited demographic and behavioral information (e.g., age, gender of sex partners in previous 12 months) was collected about each participant.

Participants in Atlanta had previously enrolled in a research study at Emory University (Sullivan et al. 2014) using venue-based time-space sampling (Muhib et al. 2001). Participants who indicated they were willing to be contacted for participation in future studies were contacted by email to complete an eligibility survey. Those meeting the eligibility criteria were invited to participate in one of the focus groups. Participants in Minneapolis were recruited via social media posts, internet advertisements, and a snowball method in which participants referred other potential participants to the study (Heckathorn 1997). Informed consent was obtained from all participants prior to participation in the FGD. This research was approved by the Emory University Institutional Review Board (IRB00046867).

Focus group discussions followed a semi-structured interview guide and lasted approximately 45–60 minutes. FGDs were facilitated by one of the authors (JJ, RB, or PSS). The facilitator followed a discussion guide that prompted questions about interactions with healthcare providers, type of insurance coverage, disclosure of sexual behavior to healthcare providers, and opinions about the tablet-based risk assessment.

At the start of each discussion, participants described their interactions with the healthcare system, including their insurance status; how frequently they discussed their sexual orientation and/or sexual behaviors with their healthcare providers; and their level of comfort in these discussions with healthcare providers. Next, the iPad-based risk assessment tool was demonstrated for the group, and each individual was given a chance to complete the assessment using fictitious data of their choosing. The risk assessment consisted of nine questions regarding recent sexual behavior and three geographical questions (see the ‘Risk assessment items’ subsection). These questions were designed to obtain the zip codes of the participant, his most recent sex partner, and the place where they most recently had sex. HIV risk depends on an individual’s risk behaviors as well as the prevalence of HIV in their potential pool of sex partners. Thus, zip code data collection in an app would enable incorporation of local prevalence data with behavioral data to calculate a modified risk score. A discussion of participants’ opinions about the usability of the tablet-based assessment tool as well as their perceptions about completing a similar assessment in their healthcare providers’ office followed.

All FGDs were recorded using digital audio recorders. After each FGD, files were transferred to a computer and transcribed by one of the researchers (JJ) who was in attendance at each FGD.

The iPad was programmed to administer a 12-item risk assessment with skip patterns. The items in the risk assessment were as follows:

Have you had sex in the past six months? [If yes, continue; if no, stop.]

All FGD transcriptions were uploaded into NVivo, a qualitative data management and analysis software (QSR International Pty Ltd. Version 8 2008). A general inductive methodology was utilized to identify frequent, salient, and significant themes in the data (Thomas 2006). First, transcripts were structurally coded according to questions in the discussion guide; transcripts were then reviewed by structural code to identify concepts across focus groups using an iterative constant comparison approach (Guest and McLellan 2003; MacQueen et al. 1998). One researcher (LT) was primarily responsible for developing the codebook and coding the transcripts. Once coding was complete, codes were reviewed by interview question in order to identify prevalent themes both across and within focus groups for each section of the interview guide.

Participant ages ranged from 20 to 46 years (median = 30 years). Overall, 14 white MSM and 10 black MSM participated in focus groups. One black participant reported having sex with men and women in the previous 12 months; the remainder reported sex with men only. The number of healthcare provider visits reported in the previous 12 months ranged from 1 to 8, with a median of 2 visits.

Most participants had discussed sexual risk behaviors and HIV testing at some point with a healthcare provider. Both patient-initiated and provider-initiated discussions of HIV risk were reported. Patients who brought up HIV during a healthcare visit themselves typically did so by requesting to be tested for HIV. Typically, if a discussion about HIV risk was initiated by a provider it was conducted via structured questionnaires or conversations. Conversations ranged from recommendations to use condoms to conversations about specific sexual practices and the emotional causes of risk behavior. These discussions were largely reported to occur only during the first office visit or on specific visits, such as one following a potential sexual exposure to HIV, but not on a routine basis. As one participant explained,

Whose doctor really does that? I mean even with a good relationship or rapport built with your doctor. Who’s really sitting around talking about sex with their doctor unless it’s a gynecologist or a urologist, possibly. But, general practitioner? I mean unless [your appointment is] really about sex. Who’s really talking to their doctor about sex? Doctors are weirded out about the stuff, too.

In addition, many participants reported they had never disclosed the gender of their sexual partners to their healthcare provider(s).

Although most participants stated they were comfortable discussing sexual risk behaviors with their providers, this was not the case for all. One participant in particular who was enrolled in a large health maintenance organization (HMO) noted that he did not feel comfortable filling out the clinic risk questionnaire provided due to the size of the practice. This participant had the perception that too many different people had access to his medical records to want to answer questions honestly. Another participant noted that he would feel singled out if his healthcare provider discussed HIV risk at every visit:

If it’s not [a new behavior]…and [my doctor] doesn’t get any indication that I’m doing any risk behavior then I don’t want to hear that [risk reduction counseling]. [If] every time I come there they’re all over me, ‘Ahhhh, you’re doing that gay thing aren’t you? Having gay sex or something. That’s what it would equal; it would equal that in my head. So, unless I was doing something risky that they were aware of I wouldn’t want to hear that. Especially not every time I came.

The degree of comfort discussing sexual risk behaviors typically depended on the degree of comfort that a participant had with their provider rather than the provider’s job title (e.g., nurse, physician, etc.). Participants reported that they would be no more or less likely to disclose sexual risk behaviors to a nurse during an initial exam than to a physician.

Opinions of the iPad-based risk assessment were generally very positive. Overall, the risk assessment was regarded as “quick and easy to do” and participants indicated a high level of willingness to complete a similar screening during a healthcare provider visit. Although opinions varied on a maximum acceptable length for the risk assessment, all agreed that the prototype tested during the FGD was an acceptable length (approximately three to five minutes in length).

Participants indicated that the app might have benefits beyond communicating their sexual risk to providers:

I think that it’s good just to start yourself thinking about what you have been doing before you go into your doctor in case you remember a question you had or it jogs your memory about something you might want to ask.

The men also suggested that they would like for the risk assessment to include other talking points to bring up with their provider because they did not always know which questions to ask. For example, participants thought it would be beneficial if the app provided information about screening tests they may need, or HIV/STD prevention programs available in the area. Many participants were also open to the idea of entering their e-mail address so that they could receive personalized sexual risk management information based on their answers to the assessment. Some men noted that receiving the assessment by e-mail would allow them to privately review the information after their visit.

Participants suggested adding questions to the tablet-based tool regarding the following topics: alcohol and drug use; relationship status with sex partner; demographic characteristics of partner; information on where the patient meets his sex partners; type of sex engaged in with different partners (e.g., oral, anal receptive, anal insertive); and use of condoms and/or lubricant.

Most participants either preferred the iPad-based method to a paper and pencil questionnaire, or expressed no preference between the two methods. The iPad-based risk assessment was perceived to be more private and secure, environmentally friendly, and more accessible for younger patients. In contrast, an older participant worried that patients his age and older might be distrustful of the technology and hesitant to use it, preferring instead to talk face-to-face with their provider.

Most participants reported that they would prefer to self-administer the risk assessment in the waiting room rather than have a provider ask the questions. These participants felt this would make it easier to honestly disclose their sexual behavior. One participant said,

I would rather [fill it out myself] than talk to [the doctor] if I didn’t know him. It would cut down my nervousness on having to see my doctor’s face if the reaction was not going to be good.

Another participant noted:

I would rather fill it out. I remember when I started going to the testing clinics, you don’t really know these people, so I know I used to tell a few stories.

The idea of the provider administering the risk assessment seemed puzzling to many participants. Participants indicated that they would be confused about why the provider needed an iPad to ask that series of questions. In addition, the amount of time available with a physician was perceived to be limited, so as one participant said:

I [would] want to fill it out before [I see the doctor], to take up time in the waiting room… If I’m with the doctor I’d rather just tell him rather than sitting here with an iPad while I’m trying to talk to him.

Filling out the risk assessment themselves was perceived to speed up the visit and maximize already limited time available for the visit. However, some participants were worried that the provider would not look at the results until they were in the examination room, even if the patient filled it out in advance.

All participants reported that an iPad-based risk assessment would result in responses to sensitive questions that were just as honest as or more honest than the same questions being asked face-to-face. Additionally, there was a strong perception that this tool would facilitate difficult conversations with the provider about HIV risk and sexual behavior. Although participants felt the iPad-based tool would not negatively impact patients’ honesty, the men did identify some barriers to using the tool accurately and effectively. Participants viewed the geographical questions as excessively intrusive and difficult to answer by all participants. Most suggested they would have no way of knowing anyone’s zip code other than their own.

Other barriers emerged after the moderator explained PrEP and the purpose of the risk assessment tool. Many participants were concerned that there would be “PrEP-seekers” who would lie about their behavior in order to try to control the recommendations made by the risk assessment. One participant echoed the sentiment of many saying,

If there was anything that let you know it was being used that way it could change how people answer. Because someone could be like, ‘oh, well, I want to try to get those meds reduced and get them at a better price. Maybe if I put answers to really make myself look risky then I might qualify for some.’

Some participants were put off at the idea of being handed a risk assessment that they felt singled them out as an MSM at their healthcare provider’s office. Particularly among black participants, there was a perception that if they, but not all participants, were handed a risk assessment that asked about engaging in sex with other men then they would feel differentially targeted compared to the patients perceived to be heterosexual. Other participants, however, disagreed and felt that their provider would be looking out for their best interests.