Acceptability of a Computer-Tailored Safer Sex Intervention for Heterosexually Active African Americans Attending an STI Clinic

Abstract

Since African Americans are disproportionately affected by HIV/AIDS, interventions that increase correct and consistent condom use are urgently needed. We report baseline acceptability data from a randomized controlled trial (RCT) testing the Tailored Information Program for Safer Sex (TIPSS), a computer-tailored intervention designed to increase correct and consistent condom use among low income, heterosexually active African Americans attending an urban sexually transmitted infection (STI) clinic. We enrolled 274 participants at baseline in an RCT – 147 in the intervention group. The intervention had high acceptability, with a mean acceptability of 4.35 on a 5-point scale. We conducted a multiple regression analysis examining demographic, structural, and sexual risk characteristics that revealed only sex to be significantly (p<.01) associated with intervention acceptability. While women were more likely than men to find the intervention acceptable, overall the results indicated broad acceptability of this intervention to the target audience. eHealth interventions are a viable option for HIV prevention among African Americans visiting a publicly-funded STI clinic. We discuss implications of these results for the future application of such programs.

INTRODUCTION

African Americans are disproportionately impacted by HIV/AIDS in the United States and make up approximately 44% of new HIV infections (Centers for Disease Control and Prevention, 2015). Their rate of HIV prevalence is eight times greater than their white counterparts (Centers for Disease Control and Prevention, 2011). Heterosexual contact is a main source of transmission among African Americans and accounted for the transmission of 34% of HIV diagnoses in 2015 (Centers for Disease Control and Prevention, 2017). Given these troubling figures, the Centers for Disease Control and Prevention (CDC) has called HIV/AIDS among African-Americans a “major health crisis” and issued a “heightened national response” to this crisis (Centers for Disease Control and Prevention, 2007). Given the severity of the epidemic among African Americans, novel intervention strategies are needed (Centers for Disease Control and Prevention, 2007; Holtgrave, McGuire, & Milan, 2007).

While efficacious interventions to reduce sexual risk behaviors among heterosexual African Americans exist (Crepaz et al., 2009; Johnson et al., 2009), they are limited insofar as most have been delivered by a health educator, counselor, or other human facilitator. Given that many community and clinic-based settings are resource constrained, consistent delivery of prevention interventions in settings such as busy sexually transmitted infection (STI) clinics has proven to be a major challenge (Rietmeijer, 2007; Ward, 2007).

Computer Technology-Based Interventions

A promising avenue for HIV behavioral intervention research is computer technology-based interventions. Computer technology-based interventions are those programs that use computer technology as the primary or sole medium to deliver an intervention (Noar, Black, & Pierce, 2009). A significant advantage of such interventions is the fact that, once developed, the cost of implementing these interventions is likely to be minimal as compared with human-delivered interventions, thus potentially facilitating their dissemination. Other advantages include the following facts: intervention fidelity is maintained through the standardization of content; computerized interventions can individually tailor intervention content; computer technologies include engaging user features such as interactivity and multimedia; and computerized interventions are flexible in terms of dissemination channels. Further, opportunities to apply new technologies to HIV prevention will only grow in the future (Bull, 2008; Noar & Harrington, 2012; Noar & Willoughby, 2012), including among disparity populations such as African Americans, due to the continued proliferation of technology (PEW Research Center, 2017).

Tailored Information Program for Safer Sex (TIPSS)

Given the promise in this area, we developed a Tailored Information Program for Safer Sex (TIPSS), a computer-tailored intervention designed to increase both correct and consistent condom use. The targeted population was low income, urban, heterosexually active African Americans attending an STI clinic. The primary theoretical basis for the intervention was the Attitude-Social Influence-Efficacy (ASE) model (de Vries & Brug, 1999; de Vries, Dijkstra, & Kuhlman, 1988; de Vries & Mudde, 1998). This theory is an integration of the Theory of Reasoned Action (Fishbein & Ajzen, 1975), Social Cognitive Theory (Bandura, 1986), and the Transtheoretical Model (Prochaska, DiClemente, & Norcross, 1992). Given the importance of skills to HIV prevention behaviors (Noar, 2008), our theoretical model also integrated condom negotiation (Noar, Carlyle, & Cole, 2006; Noar, Morokoff, & Harlow, 2002) and skills training principles (Segrin & Givertz, 2003).

The TIPSS program assesses individuals on ASE variables (condom stage of change, attitudes, norms, self-efficacy, and negotiation) and provides tailored feedback based upon individuals’ scores on these variables. The program branches based upon whether individuals have main, main and casual, or only casual sexual partners. Skills training exercises provide learning opportunities for both correct condom use and condom negotiation skills. The program was developed through extensive formative research with the target audience, including quantitative survey research (Noar, Crosby, Benac, Snow, & Troutman, 2011a) and qualitative focus groups (Noar et al., 2012). A detailed description of the intervention itself is available elsewhere (Noar et al., 2011b).

We tested the efficacy of the TIPSS program in a 2-arm randomized controlled trial (RCT). Here, we report on baseline data from the trial, in particular focusing on the acceptability of the TIPSS program among the target population. Understanding whether these types of eHealth programs are acceptable to at-risk audiences is important, and is a likely prerequisite to efficacy. Additionally, while the specific technologies on which eHealth and mHealth interventions are developed may rapidly change and advance, understanding how interventions and intervention components may impact program acceptability and success can be useful for the development of future interventions, regardless of technology platform.

Method

Participants

We conducted the study from February 2010 to May 2012, and completed baseline data collection in February, 2012. We recruited study participants from a publicly funded, urban STI clinic in a city in the Southeastern United States. Our recruiter approached all potential participants (patients of the clinic) in the clinic waiting room and briefly described the study. Those interested in participating were led to a private room (adjacent to the clinic) for screening, which took place using a 1-page paper form. Inclusion criteria were as follows: 1) patient of the clinic in the past 7 days (including today); 2) identified as Black or African American; 3) aged 18 – 29 years; 4) heterosexually active in the past 3 months (i.e., vaginal sex with an opposite sex partner); 5) not (knowingly) IV positive; 6) not pregnant or intending on becoming pregnant (for women) or impregnating main partner (men) in the next 3 months; and 7) not enrolled in another condom study at the clinic in the past 3 months. Participants who identified as homosexual were not excluded from the study unless they could not report having engaged in heterosexual activity in the past 3 months.

Of the 432 potentially eligible participants that we screened, 140 did not meet inclusion criteria and 18 met criteria but declined to participate in the study. Thus, we enrolled a total of 274 participants. We obtained written informed consent from all participants before they began the study.

Study Design

Our study was a 2-arm randomized controlled trial, with concealment of allocation techniques used to minimize allocation bias. A random number generator built into the computer software program, Filemaker Pro 10.0, randomly assigned each individual to an intervention group (treatment or control). This ensured that each participant had an equal chance of being randomized to the treatment or control condition. After the baseline visit, a 3-month in-person visit was scheduled for follow-up assessment and provision of condoms.

Intervention Group

We provided the intervention group the TIPSS intervention (Noar et al., 2011b), which was delivered locally on a laptop computer using Audio Computer Assisted Survey Instrument (ACASI) technology, in a private room. The program ran on Filemaker Pro 10.0 software, in which pre-programmed algorithms governed what tailored feedback was given to participants, based on their responses to the computerized assessment. Skills training exercises were also provided, focusing on correct condom use (for all intervention participants) and condom negotiation (specific to main and/or casual partners). To be sensitive to low literacy individuals, all questions and feedback used audio-assisted narration. The intervention took participants 30–45 min to complete. Once complete, participants were given a stage of change-matched paper report, up to 12 packets of lube, and the opportunity to select up to 36 condoms from a selection of 11 different types. Intervention participants also received a study incentive of $35 at baseline and $50 at follow-up.

Control Group

We provided the control group a similar computerized assessment as the intervention group, but they received no tailored feedback or skills training exercises. The assessment took 20–30 min to complete. When the assessment was complete, control group participants were given an information-only safer sex pamphlet, up to 12 packets of lube, and 12 Lifestyles extra strength condoms (the number and type of condoms that were available to patients at the clinic). Control participants also received the study incentive of $35 at baseline and $50 at follow-up.

Measures

We used a computerized questionnaire to assess demographics and structural factors, ASE model variables, substance use, sexual history and condom use, and in the case of the intervention group, acceptability of the intervention. Specific measures are described below.

Demographic and structural factors.

Demographic characteristics were sex, race/ethnicity, age, education and income. Structural factors were incarceration history and unstable housing. We assessed incarceration history by asking whether participants had ever been incarcerated in a detention center or prison. We assessed unstable housing by asking participants how many different places they had lived over the past 6 months.

Number of sexual partners.

We assessed number of sex partners by asking participants how many male or female sexual partners they had had in the past 3 months. This included both opposite sex and same sex partners.

Condom use.

We assessed condom use by asking participants how many times they did and did not use condoms when having vaginal or anal sexual intercourse (assessed separately, but combined for analysis) with their main/casual partner(s) over the past 3 months. Condom-protected occasions were then divided by the total number of intercourse occasions, yielding a ratio from 0 to 1.0.

Stage of change for consistent condom use.

We assessed stage of change for consistent condom use during vaginal and anal sex with a 4-item algorithm adapted from a previous study (Grimley, Prochaska, Velicer, & Prochaska, 1995). The items assess behavioral, intentional, and temporal criteria for consistent condom use with a main/casual partner. Figure 1 illustrates how individuals are classified into the stages of change for consistent condom use.

Figure 1.

Visual depiction of algorithm for classifying individuals into stages of change for consistent condom use.

Sexually transmitted infections.

We assessed STIs by asking if the participant had had an STI (i.e., chlamydia, herpes, syphilis, gonorrhea, HPV, but not including HIV) in the past 6 months. Responses were yes, no, and I might have (don’t know).

Intervention acceptability.

We assessed acceptability (intervention group only) with 8 items derived from previous research (Brown-Peterside, Redding, Ren, & Koblin, 2000). These items (see Table 1) assessed whether participants liked the intervention, if it kept their attention, how much information they read or listened to, whether it fit their needs, whether they felt it was designed for them, how easy or difficult they found the program to be, whether they would come back for another session, and whether they would recommend the program to a friend. All items used 5-point Likert scales. Our factor analysis (described below in results) found these items to form a reliable scale. Coefficient alpha of this acceptability scale was .84.

Table 1.

TIPSS Acceptability Items

Item	Response scale	Mean	SD	Factor loading
How much did you like the TIPSS computer program?	Not at all -Very much	4.22	.88	.81
Did the TIPSS program keep your attention?	Not at all -Very much	4.28	.89	.74
How much of the information provided in the TIPSS program did you read or listen to?	None -All of it	4.57	.73	.63
How well did the information provided in the TIPSS program “fit” your needs?	Definitely no -Definitely yes	4.10	.94	.72
Did you feel that the information in the TIPSS program was designed specifically for you?	Definitely no -Definitely yes	3.62	1.18	.71
How difficult or easy did you find the TIPSS computer program to be?	Very difficult -Very easy	4.50	.77	.52
Would you come back for another session with the TIPSS computer program?	Definitely no -Definitely yes	4.77	.55	.74
Would you recommend the TIPSS computer program to a friend?	Definitely no - Definitely yes	4.72	.61	.71

Note. All items were asked on 5-point response scales.

Analysis

We first examined descriptive data for the study. We then ran a factor analysis on the acceptability scale items before running a hierarchical multiple regression to determine whether acceptability differed across demographic, structural, and/or sexual risk characteristics. We describe each analysis in more detail directly before reporting the results.

Results

Characterizing the Sample

Since the current article focuses only on acceptability of the intervention group, we describe and focus on characteristics of only that group in Table 2 (though statistical comparisons of all variables in Table 2, using t tests and chi-squares, revealed no significant (p<.05) differences compared to the control group). As can be seen, the intervention sample was slightly more female (59%) than male, and mean age was 23.5 years. Many participants had a high school diploma (45%) or more education (31%), and most were unmarried (82%). Most participants were unemployed (51%) and the sample was largely low income (72% made less than $10,000 in the past year). More than half of the sample (52%) had been incarcerated in their lifetime.

Table 2.

Baseline Demographic and Sexual Risk Characteristics

Variable	Intervention Group (N=147)
	n	%	Mean	Median
Sex
Male	60	41	-	-
Female	87	59	-	-
Age - Mean (SD)	-	-	23.5	23.0
Hispanic/Latino	3	2	-	-
Education level
Less than HS diploma	35	24	-	-
HS diploma	67	45	-	-
2 year degree	40	27	-	-
4 year degree	1	1	-	-
School beyond college	4	3	-	-
Marital status
Unmarried	120	82	-	-
Unmarried living with partner	22	15	-	-
Married	5	3	-	-
Income - past year
0	24	16	-	-
1.00–5,000	43	30	-	-
5,001–10,000	38	26	-	-
10,001–30,000	36	25	-	-
More than 30,001	4	3	-	-
Employment
No	75	51	-	-
Yes, part time	40	27	-	-
Yes, full time	32	22	-	-
Ever incarcerated
Yes	76	52	-	-
No	71	48	-	-
Unstable housing
No. of places lived (past 6 mos)	-	-	1.56	1.0
Sexual partner types
Main only	56	38	-	-
Main and casual	61	42	-	-
Casual only	29	20	-	-
# of sexual partners
# of sex partners (past 3 mos)	-	-	2.76	2
# of sex partners (past year)	-	-	4.89	3
STI in past 6 months
No	78	54	-	-
Yes	59	41	-	-
I might have (don’t know)	7	5	-	-
Visited clinic in past 6 months
Not at all	69	48
Once	45	31
Twice	25	17
Three or more times	5	4
Provided sex in exchange for something in past 6 months
Yes	15	10	-	-
No	130	90	-
Tested for HIV in past 6 months
Yes	82	57	-	-
No	63	43	-	-
Past 3 months used alcohol to get drunk
Never	47	32	-	-
1 –2 times per month	61	42	-	-
1–2 times per week	23	16	-	-
About 3 times per week	10	7	-	-
Almost every day	4	3	-	-
Past 3 months used marijuana
Never	55	39	-	-
1–2 times per month	32	23	-	-
1–2 times per week	6	4	-	-
More than 1–2 times per week	8	6	-	-
Almost every day	40	28	-	-
Proportion of condom use
Main partners	-	-	.30	.36
Casual partners	-	-	.61	.38

Most participants had either a main sexual partner only (38%) or a main partner and additional casual partners (42%). Twenty percent had causal partner(s) only. Median number of sex partners in the past year was 3, and 41% of the sample reported having an STI in the past 6 months, with 5% reporting they may have had an STI. Proportion of condom use in the past 3 months was 30% with main partners and 61% with casual partners.

Classification of individuals into condom stages of change for main partners (n=118) and casual partners (n=90) revealed the following stage distributions for main and casual partners, respectively: precontemplation (21%, 4%), contemplation (59%, 38%), preparation (10%, 31%), action/maintenance (10%, 27%). These data reveal variability with regard to individuals’ readiness to use condoms consistently with main and casual sexual partners. They also reveal that individuals were in more advanced stages of change with regard to causal as compared to main partners (see Figure 2).

Figure 2.

Stages of change for consistent condom use with main and casual sexual partners.

Acceptability of the Intervention

Table 2 displays the means and standard deviations of each of the individual acceptability items. The following acceptability indicators were rated most highly: coming back for another session (M=4.77), recommending it to a friend (M=4.72), reading or listening to the information/advice (M=4.57), and finding the program easy to use (M=4.50). The lowest rated item, and the only item rated below 4.0, was with regard to whether participants felt the information was designed specifically for them (M=3.62).

To assess whether the acceptability items could be formed into a scale, we computed a maximum likelihood factor analysis with all 8 items. Eigenvalues >1 indicated there were possibly two factors, but all items loaded onto the first factor (with only a single item, on ease of use of TIPSS, loading on both factors). The scree plot indicated only one factor. Evidence suggests that the scree plot may be more reliable given that the eigenvalue >1 rule has been shown to over-extract factors (Zwick & Velicer, 1986). Thus, we forced all items into a single factor, producing a solution that accounted for 49.26% of the variance. Factor loadings were all acceptable at >.50, and are listed in Table 2. The overall acceptability of the intervention, as assessed by this 8-item scale, was M=4.35 (SD = .58).

Differences in Acceptability

In order to assess whether acceptability differed across key demographic, structural, and sexual risk characteristics, we conducted a hierarchical multiple regression analysis. In step 1, we entered the demographic characteristics of sex, age, and education. In step 2, we added the structural factors of incarceration history and unstable housing. In step 3, we entered condom use, number of sexual partners in the past three months, and reported STI in the past 6 months.

At step 1, the model was statistically significant, F(3, 139) = 3.55, p=.016, multiple R=.27. The model remained significant in step 2, F(5, 139) = 2.37, p=.042, multiple R=.29, and step 3, F(8, 139) = 2.10, p=.041, multiple R=.34. However, the change in R was not statistically significant (p>.05) at steps 2 or 3. This indicated that step 2 and 3 variable(s) were not significantly associated with intervention acceptability over and above step 1 variables.

The key significant predictor in the model was sex, with females reporting higher levels of intervention acceptability (M=4.46, SD=.50) than males (M=4.19, SD=.64). Number of sexual partners approached significance (p=.06), with a trend toward people who had fewer sexual partners reporting greater intervention acceptability. No other associations were statistically significant (see Table 3).

Table 3.

Predictors of TIPSS Acceptability

Predictor Variables	β	r	R2	ΔR2
Step 1		.27*	.07	.07*
Sex	.259**
Age	.151
Education	−.018
Step 2		.29*	.08	.01
Sex	.275**
Age	.124
Education	−.008
Incarceration history	.090
Unstable housing	.036
Step 3		.34*	.11	.03
Sex	.247**
Age	.103
Education	.100
Incarceration history	.090
Unstable housing	.031
Condom use	.064
Multiple partners	−.162
STI in past 6 months	−.035

^*p<.05.

^**p<.01.

^***p<.001.

Note. Sex was coded 1=male, 2=female; education was coded on a 1–6 categorical scale, with 1=9^th grade or less and 6=some graduate school/graduate degree; incarceration was coded 1=no, 2=yes; multiple partners was coded as 1=1 partner in the last three months and 2=2 or more partners in the last three months; STI was coded as 1=no STI in the past 6 months and 2=STI in the past 6 months.

Discussion

As computer and Internet technologies continue to develop and proliferate, health communicators have increased opportunities for computer-delivered health education and health promotion (Noar & Harrington, 2012). Indeed, in the HIV/STI prevention domain, reviews reveal an increase in the number and type of eHealth interventions being developed and tested (Jones & Salazar, 2016; Lim, Hocking, Hellard, & Aitken, 2008; Noar, 2011; Noar et al., 2009; Noar & Willoughby, 2012). To date, however, few of these applications have been directed toward heterosexually active African Americans, and few applications have been tested in STI clinics. While technology continues to adapt, understanding how certain elements of technology-based interventions impact acceptability by a target audience provides valuable insights for health communicators and program developers, even as delivery platforms change and evolve. To that end, we report on the application of a computer-tailored intervention to a critical at-risk audience – low income, heterosexually active African Americans attending an STI clinic. We used baseline data from our RCT to examine 1) characteristics of the sample we reached, 2) overall intervention acceptability, and 3) factors associated with intervention acceptability.

Results indicated that we reached a low income, high-risk, and underserved audience. This population exhibited high rates of unemployment (>50%), incarceration (>50%), STIs (>40%), and trading sex for money or goods (10%). Condom use was also fairly low, particularly with main sexual partners, despite the fact that greater than 40% of the sample had casual partner(s) in addition to their main partner.

Results indicated high acceptability of the computer-tailored intervention, with a mean acceptability score of 4.35 on a 5-point scale. Participants reported that they engaged with the program, paid attention to its content, and found it easy to use. Perhaps most promising, they almost uniformly reported they would come back for another session and recommend the program to a friend. This high level of acceptability bodes well for the future application of computer-based interventions for HIV/AIDS prevention among this population. It is also worth noting that these data dovetail with qualitative data collected during intervention development, which indicated that participants were very engaged in the topic of sexual health (Noar et al., 2012). Anecdotally, we found that participants desired additional opportunities to discuss and address the topic of relationships, sexual health and STIs.

In addition to overall acceptability, we examined participant, structural, and sexual factors as possible predictors of intervention acceptability. The purpose was to examine whether the intervention was more or less acceptable to particular subgroups. In regression analyses, with all variables entered, the only variable that was statistically significant was sex, in that women found the intervention more acceptable than did men. Why was this the case? Multiple studies find men to have more negative attitudes toward condoms than women (Campbell, Peplau, & Debro, 1992; Sacco, Levine, Reed, & Thompson, 1991). Thus, it may be that an intervention suggesting that individuals use condoms more consistently was found to be more acceptable to women. Another related explanation relates to the fact that some research finds men to be more critical of sexual health promotion messages compared to women (Noar, Palmgreen, Zimmerman, Lustria, & Lu, 2010; Van Stee et al., 2012). Thus, women may simply be more receptive to sexual health messages of any type than men. It is important to note however, that mean acceptability was >4 even for men, and thus the intervention was clearly acceptable to both groups. Other than sex, no other variables were statistically significant, suggesting that the intervention was broadly acceptable to this audience. The finding that a tailored eHealth intervention was acceptable to participants also supports research conducted on other platforms. In a meta-analysis of text message based interventions, Head and colleagues (2013) found that tailored and targeted interventions were most efficacious.

Conclusion and Future Directions for eHealth HIV Prevention Programs

We found that a computer-tailored intervention to increase correct and consistent condom use was broadly acceptable to low income, heterosexually active African Americans attending an STI clinic. This high level of acceptability bodes well for the future application of computer-based interventions for HIV/AIDS prevention among this population. Future eHealth HIV prevention programs might include cell phone interventions, social media, and games for health (Bull, Levine, Black, Schmiege, & Santelli, 2012; Lim et al., 2008; Noar & Willoughby, 2012). Indeed, with data indicating that the majority of African Americans possess and are heavy users of cell phones (PEW Research Center, 2017), mobile interventions and applications are a particularly promising future direction. Text message applications can be designed to operate on their own (Levine, McCright, Dobkin, Woodruff, & Klausner, 2008) or in concert with TIPSS or other computer/web-based programs (Muessig et al., 2013), and African Americans have indicated that sexual health text message interventions are of interest to them (Wright, Fortune, Juzang, & Bull, 2011). In one study, text messaging was found to serve as a way to reach out to African American adolescents with information supporting an existing HIV prevention intervention (Cornelius & St. Lawrence, 2009). Preliminary evaluation work conducted on another program found that young African American men who received HIV prevention text messages for 12 weeks showed increased monogamy at follow-up compared to a control group (Juzang, Fortune, Black, Wright, & Bull, 2011). Thus, mobile technologies offer a promising and potentially efficacious mode of delivering sexual health messages to African American populations at risk of HIV infection.