Using Hierarchical Regression to Examine the Predictors of Sexual Risk-Taking Attitudes among Adolescents Living with Human Immunodeficiency Virus in Uganda

Abstract

Purpose:

We explored the factors influencing sexual risk-taking attitudes—defined as beliefs and values regarding sexual activity—among adolescents living with human immunodeficiency virus (ALHIV) in Uganda.

Methods:

The study used baseline data from a five-year cluster-randomized control trial (2012–2018) among 702 ALHIV in Uganda. Participants were aged 10–16 years, HIV-positive, taking antiretroviral therapy, and living within a family. We fitted hierarchical regression models to assess the demographic, economic, psychological, and social predictors of sexual risk-taking attitudes. Using R², the final model explained 11.4% of the total variance.

Results:

Under economic factors, caregiver being formally employed (β = −0.08, 95% confidence interval [CI]: −0.10–0.06, p < .001), and the ALHIV working for pay (β = 1.78, 95% CI: 0.28–3.29, p = .022), were associated with sexual risk-taking attitudes. Among the psychological factors, more depressive symptoms (β = 0.22, 95% CI: 0.11–0.32, p < .001) were associated with more approving attitudes toward sexual risk-taking. Family and social factors including communicating with the caregiver about HIV (β = 1.32, 95% CI: 0.56–2.08, p = .001), sex (β = 1.09, 95% CI: 0.20–1.97, p = .017), and experiencing peer pressure (β = 3.37, 95% CI: 1.85–4.89, p < .001) were also associated with more approving attitudes toward sexual risk-taking. The final model explained 11.54% of the total variance.

Discussion:

Economic, psychological, and social factors influence sexual risk-taking attitudes among ALHIV. There is a need for more research to understand why discussing sex with caregivers improves adolescents’ positive attitudes toward sexual risk-taking. These findings have significant ramifications in preventing sexual transmission of HIV among adolescents in low-income settings.

Sub-Saharan Africa (SSA) accounts for 90% of all global human immunodeficiency virus (HIV) infections among adolescents aged 10–19 years [1]. While major strides have been made in reversing the trends in HIV morbidity and mortality in other age groups, the progress in adolescents is still lagging. For instance, adolescents are the only age group in which HIV-related deaths have not reduced in the last decade, currently accounting for one in every six HIV-related deaths [2]. The 2021 Joint United Nations Program on HIV/AIDS report indicated that 37% of all new HIV infections in Uganda were among youths aged 15–24 years [3]. While many adolescents acquire HIV perinatally, the majority acquire it sexually. The adolescents who acquired HIV perinatally face different challenges compared to those who get infected later in life [4]. For example, adolescents who acquire HIV sexually may have higher rates of risky sexual behaviors and may be at an increased risk of transmitting HIV to their partners [5].

Adolescence is a period of transition from childhood to adulthood. Adolescents undergo biological, psychological, and behavioral changes, which render their decision-making more reactionary and oriented toward satisfying their desires, including sexual curiosity [6]. Therefore, irrespective of their HIV status, adolescents are more likely to engage in risky sexual behaviors, such as early sexual exploration and unprotected sex [7,8]. Such behaviors may result in acquiring or spreading HIV and sexually transmitted infections (STIs) [7,9]. Most research on sexual risk-taking behaviors is among HIV-negative adolescents, with only limited literature on adolescents living with HIV (ALHIV) [10-12]. This research gap is surprising because risky sexual behaviors such as unprotected sex among ALHIV carry a higher risk of HIV transmission since these adolescents are HIV-infected. Therefore, addressing risky sexual behaviors in ALHIV is crucial in controlling HIV spread. Furthermore, there is low utility of existing risk scores—such as the HIV risk score [13]—in studies conducted in SSA settings, with various studies defining sexual risk-taking differently [8].

Sexual risk-taking is driven by multidimensional factors. Toska et al. (2017) conducted a systematic review on sexual risk-taking behaviors among ALHIV using 35 studies from 13 SSA countries [4]. They found that individual factors including participants’ increasing age, female gender, low STI prevention knowledge and attitudes, and structural factors such as poverty, school dropout, unemployment, and lack of social support were all associated with increased sexual risk-taking behaviors. Other studies showed that cohesive families are more likely to communicate with adolescents about risky sexual activities, which protects the adolescents from these behaviors [14,15]. Consequently, adolescents from broken families—for instance the orphaned ALHIV—may face a higher risk of engaging in risky sexual behavior [16,17]. These results are essential because many HIV-positive adolescents have lost their biological parents due to HIV.

While social factors have been extensively studied, research on the role of economic and psychological factors in influencing sexual risk-taking behaviors is still warranted, especially among ALHIV in low-income settings. The few studies that examined economic factors found poverty to precipitate sexual risk behaviors [18]. A study among young women in Zimbabwe found that lower social economic status was associated with early marriage, early sexual debut, and engaging in transactional sex [19]. Challenges like food insecurity and the need for financial support may drive adolescents to indulge in transactional sexual activities [18,19]. Moreover, recent intervention studies in SSA showed that economic empowerment interventions were protective against sexual risk-taking among adolescents [20-23], which implies that poverty is an underlying factor in driving risky sexual behaviors among adolescents.

This article presents findings from a study conducted in Southern Uganda, where we examined the individual and family, economic, psychological, and social factors associated with sexual risk-taking attitudes among ALHIV.

Theoretical framework

Reducing the horizontal transmission of HIV among ALHIV involves understanding the drivers of sexual risk-taking behaviors in this population. This study is guided by Bronfenbrenner’s socioecological model (1977), a theory-based framework that examines multilevel factors influencing individual behavior [24]. As per Bronfenbrenner, understanding human behavior involves examining multiple systems of influence. This model has been used in the present study to better understand the range of factors in the adolescents’ lives that place them at increased risk of engaging in sexual risk behaviors at both the microlevel and mesolevel. In our study, the microlevel comprised the individual factors including adolescents’ demographic characteristics such as gender, age, orphanhood status, and school enrolment status; economic factors including household asset ownership, caregiver being formally employed, and whether adolescents worked for pay; and psychological factors including adolescents’ HIV status disclosure, HIV stigma, and depression. These microlevel factors facilitate interactions between the individual and their immediate environment [24]. We included family and social factors such as peer pressure and adolescent-caregiver communication on HIV, sex, and puberty at the mesosystem level to explain the individual’s interactions with their immediate environment. The social support system consists of the network of individuals and support that the adolescents receive which either protects or puts them at risk of engaging in risky behaviors.

Methods

Study design

This study used baseline data from 702 ALHIV enrolled in a cluster-randomized controlled trial (2012–2018). The main aim of the parent study was to examine the impact of a novel family-level economic empowerment intervention on adherence to antiretroviral therapy among ALHIV. Adolescents were included in our study if they were aged 10–16 years, HIV-positive after confirmation using the medical report, were prescribed to take antiretroviral therapy and were living with a family. Details are elaborated in the study protocol [25].

Study setting

Study participants were recruited from 39 clinics in Southern Uganda, a region heavily impacted by HIV/AIDs. It was in this region that the first cases of HIV were identified in Uganda in the early 1980s [26]. As per the Uganda Aids Commission report, the HIV prevalence in Masaka region is 11.7%, which is twice higher than the national average of 5.4% [27]. The region is predominantly rural, with most households are struggling with poverty [28].

Measurements

Trained research assistants collected data through an interviewer-administered questionnaire. Interviews lasted an average of 90 minutes each and were conducted in Luganda, the most widely used language in the study area. The interviewers were all fluent in both English and Luganda and data collection tools were translated from English to Luganda and back to English by a team of experts from Makerere University Institute of languages. We collected demographic data, including participants’ gender, age, school attendance, and orphanhood status (both parent alive, one parent alive or duo orphan), and data on the following variables.

Sexual risk-taking attitudes

The primary outcome of interest for this study was sexual risk-taking attitudes. Many studies acknowledge the challenges in measuring sexual risking-taking behaviors among adolescents [20,29,30] and ours was no exception. Although we assessed for sexual risk-taking behavior using the question "Have you ever had sexual intercourse?", only a small number of adolescents (n = 33) reported having sexual intercourse. This number was too small to generate meaningful inferential analyses on the outcome. This low number could be attributed to a high age at first sexual intercourse among Ugandans. The median age at coitarche in Uganda was reported at 17.1 years in the most recent Uganda Demographic Health Survey [31], while the median age of adolescents in our study was 12.4 years. Previous research has shown that sexual risk-taking attitudes closely predict sexual risk-taking behavior [32]. Several studies have used sexual risk-taking attitudes to measure sexual risk-taking behaviors [20,21,30]. Hence, we used sexual risk-taking attitudes as a proxy for ALHIV future propensity to participate in risky sexual behaviors. Sexual risk-taking attitudes were measured using a five-item Likert-based scale that ranged from "1 = Never" to "5 = always" (alpha = 0.83). The items were added up to create a total score for sexual risk-taking attitudes (range 5–25). Higher scores suggested a more approving attitude toward sexual risk-taking behaviors. Specifically, the adolescents were asked to give their opinion on the following items on the scale.

1. "It is okay to have sex with someone you have just met."

2. "It is okay to have sex with someone they love."

3. "It is okay to have sex before marriage."

4. "It is okay to force a girlfriend/boyfriend to have sex even when they do not want to."

5. "It is okay to have sex without protection with someone you know."

The scale has been in previous studies [20,30].

Economic factors

We collected information about household asset ownership, whether the head of the household had formal employment, and whether the participant was doing any job for a payment. To measure asset ownership, we used a 20-item asset index, which asked the adolescents’ families whether they owned household assets such as a house, land, and other property. From this index, we generated a scale that ranged from 0 to 20, with a higher score denoting more asset ownership. The scale was dichotomized into 0 = few assets (if participants scored less than 13) and 1 = more assets (if the participants scored 13 or more). This categorization was based on the 60th percentile as recommended by the world bank [33]. In addition, the adolescents were asked whether they were involved in work for pay, had any money saved, and whether the caretaker had a formal job or was earning a wage.

Psychological factors

The psychological factors in our study included depressive symptoms, HIV status disclosure, and HIV stigma. Depressive symptoms were measured using a 14-item short version (alpha = 0.63) of the Children’s Depression Inventory instead of the 27-item version [34]. The 14-item Children’s Depression Inventory scale was validated and found to be a reliable measure of depressive symptoms [35]. It has been used in several settings, including among adolescents in low-income settings [36].

To measure HIV status disclosure, we asked the adolescents two questions. (1) “Do you keep your HIV status a secret from others, such as friends and other family members?" This item was measured on a Likert scale ranging from 1 = Never to 5 = Always. The item was reverse-coded so that a higher score indicated more disclosure and (2) "How often do you talk to people about your HIV status?" The item was on a Likert scale ranging from 1 = “Never” to 5 = “All the time.” The two items were combined to form a composite score, with a higher score indicating more disclosure. The measure has previously been used to measure HIV disclosure [37].

We used a nine-item scale to measure internalized stigma among adolescents (alpha = 0.74). The scale was adapted from the Berger stigma scale [38]. It includes statements that HIV-positive patients made about themselves, such as "When people know I have HIV, I feel uncomfortable around them." Adolescents were asked to indicate how much they agreed with each item on a four-item Likert scale (ranging from 1 = strongly disagree to 4 = strongly agree). The scale has been used to measure stigma in the same population [37].

Family level and social factors

Studies have found that parental communication, peer pressure, and household composition all influence adolescent behaviors [20,30]. Hence, we collected information on the three measures in our study. Participants were asked how often they discussed topics such as HIV, bad friends, puberty, and engaging in sex with their caregivers. Each response was on a five-item Likert scale ranging from 1 = "Never" to 5 = "Always." A higher score indicated more frequent adolescent-caregiver communication. The measure has been previously used [30]. We also determined how much pressure the participants felt from their peers to engage in sexual activities. The Likert-scale–based responses were collapsed into three options: no peer pressure, some peer pressure, and a lot of peer pressure. This measure has been previously used in the same setting [20].

Analysis.

Data were analyzed using Stata version 17.0. Descriptive statistics included summarizing categorical data using percentages and continuous data using means and standard deviations. We fitted a hierarchical regression model to assess the incremental effect of the predictors of sexual risk-taking behaviors among ALHIV. A new block of variables was added in each model to determine its effect in explaining the observed variation in the outcome. The first block of variables constituted model 1 and included the following demographic factors: gender, age, school enrolment, and orphanhood status. In model 2, economic factors were introduced, including household asset ownership, caregiver earning salary or wage, and adolescents working for pay. Psychological factors including HIV status disclosure, HIV stigma, and depressive symptoms were introduced in model 3. In model 4, we introduced the following household and social factors: household size, peer pressure, and communication with the caregiver about various topics including HIV, having sex, and puberty. To ensure suitability for linear regression, the data were checked for normality of the residuals, absence of multicollinearity, and heteroskedasticity. Specifically, we assessed for multicollinearity using the variance inflation factor test and aimed for a variance inflation factor value of less than 10. Also, using the Breusch–Pagan/Cook–Weisberg test, we tested for the absence of heteroskedasticity. We plotted normal q-q, p-p, and kernel density plots for the residuals after fitting a regression model to test the residuals’ normality. The plots indicated that there was normality. The increase in total variance explained (incremental R²) in each model was reported. The significance level was set at 0.05. During the analysis, data were declared to be survey data, as one of the means to adjust for clustering at the level of clinics. In addition, we reported robust standard errors to further cater for clustering.

Ethical consideration.

The Columbia University Institutional Review Board in the United States (#AAAK3852) and the Makerere University School of Public Health Ethics Review Committee in Uganda (#210) approved study procedures. In addition, the study was authorized by the Uganda National Council for Science and Technology (UNCST, SS2969). A written informed consent was obtained from the caretakers of the adolescents. In addition, the adolescents provided assent prior to participating in the study.

Results

Sample characteristics

We recruited 702 participants in the study, with 56% being females. The mean age was 12.4 years (range: 10–16 years). Regarding school attendance, 87% of the adolescents were in school at the time of enrollment in the study. Only 36% of the participants had both parents (nonorphans), while 26% had lost both parents (Table 1).

Table 1.

Baseline characteristics of 702 adolescents living with HIV in Uganda

Characteristics	n (%) or mean (SD)
Individual and household factors
Female gender	396 (56.4)
Mean age in completed years ± SD (Range: 10–16)	12.4 ± 1.98
Household size
Number of adults in the household	5.7 ± 2.6
Number of children below 18 years in the household	2.3 ± 1.9
Education status
Enrolled in school	613 (87.3)
Not in school	89 (12.7)
Orphanhood status
Both parents are deceased	182 (26.4)
At least one of the parents is still alive	262 (38.0)
Both parents are alive	246 (35.7)
Economic factors
Have money saved somewhere	205 (29.2)
Household caregiver is formally employed or works for a wage	75 (10.7)
Adolescent works for pay	65 (9.3)
Household asset ownership
Few assets	464 (66.1)
Many assets	238 (33.9)
Psychosocial factors
Disclosure ± SD (Range: 1–10)	5.0 ± 2.1
Stigma ± SD (Range: 9–35)	18.6 ± 5.9
Depressive score ± SD (Range: 1–22)	6.7 ± 3.5
Communicate with the caregiver about
HIV	522 (74.4)
Having sex	185 (26.4)
Bad friends	297 (42.3)
Puberty	412 (58.7)
Peer pressure
Does not feel peer pressure	571 (81.9)
Feels peer pressure sometimes	88 (12.6)
Feels peer pressure most of the time	38 (5.5)

Sexual risk-taking behaviors

Table 2 shows the sexual risk-taking behaviors of the participants. We found that 9% of the ALHIV were involved in a romantic relationship and 5% (n = 33) previously had sexual intercourse. Ten of the 33 participants (30%) who reported engaging in sexual intercourse reported having unprotected sex. The mean sexual risk-taking attitudes score was 9.4 ± 5.4.

Table 2.

Sexual risk-taking behaviors among adolescents living with HIV in Uganda

Characteristic	Number (%) or mean (SD)
Involved in an intimate/romantic relationship	60 (8.7)
Have had sexual intercourse	33 (4.7)
Method of protection during the last sexual intercourse (n = 33)
Had unprotected sex	10 (30.3)
Condom	11 (33.3)
Had sex in the last 30 days (n = 702)	16 (2.3)
Number of times participant had unprotected sex in last 30 days (n = 19)
None	8 (42.1)
Once	6 (31.6)
Twice	1 (5.3)
Thrice	1 (5.3)
Not stated	3 (15.8)
Sexual risk-taking attitudes (range: 1–5)
’s okay for people my age to have sex with someone they’ve just met (n = 693)	1.85 (1.47)
It’s okay for people my age to have sex with someone they love (n = 690)	1.85 (1.36)
It’s okay for people to have sex before marriage (n = 688)	2.00 (1.46)
It’s okay to force a partner to have sex even when they do not want to (n = 696)	1.90 (1.40)
It’s okay to have sex without protection with someone you know (n = 691)	1.89 (1.40)
Mean sexual risk-taking attitudes score ± SD (Range: 5–25)	9.4 ± 5.4

Hierarchical regression results

Results from the hierarchical regression are presented in Table 3. Model 1 included demographic factors and explained 0.70% of the total variance (R² = 0.0070). None of the demographic factors was significantly associated with sexual risk-taking attitudes. In model 2, we introduced economic factors. Factors in model 2 explained 1.59% of the total variation in sexual risk-taking attitudes (R² = 0.0159). Adding economic factors to the model increased R² by 0.0089. We found that adolescents from households where the primary caregiver had formal employment (β = −0.07, 95% confidence interval [CI]: −0.10 to −0.05, p < .001) had disapproving attitudes toward sexual risk-taking.

Table 3.

Hierarchical models for the factors associated with sexual risk-taking attitudes among ALHIV in Uganda

	Model 1	Model 2	Model 3	Model 4
Characteristics	β (95% CI)	β (95% CI)	β (95% CI)	β (95% CI)
Demographic factors
Age in completed years	0.11 (−0.10 to 0.33)	0.11 (−0.10 to 0.32)	0.18 (−0.03 to 0.39)	0.06 (−0.15 to 0.27)
Female gender	−0.75 (−1.56 to 0.05)	−0.56 (−1.38 to 0.26)	−0.33 (−1.11 to 0.44)	−0.45 (−1.20 to 0.31)
Orphanhood status
Double orphan	Ref	Ref	Ref	Ref
One parent is alive	0.13 (−1.10 to 1.37)	0.14 (−1.04 to 1.32)	0.02 (−1.16 to 1.20)	0.05 (−1.03 to 1.13)
Both parents are alive	−0.42 (−1.55 to 0.72)	−0.44 (−1.58 to 0.69)	−0.60 (−1.78 to 0.57)	−0.38 (−1.43 to 0.68)
Enrolled in school	0.22 (−1.47 to 1.91)	0.32 (−1.49 to 2.14)	0.75 (−1.06 to 2.56)	0.97 (−0.67 to 2.62)
Economic factors
Household asset ownership		−0.04 (1.49–2.14)	−0.02 (−0.16 to 0.12)	−0.03 (−0.16 to 0.11)
Caregiver is employed		−0.07 (−0.10 to −0.05)	−0.08 (−0.10 to −0.07)	−0.08 (−0.10 to −0.06)
Adolescent works for pay		1.70 (0.13 to 3.28)	1.85 (0.37 to 3.33)	1.78 (0.28 to 3.29)
Psychological factors
HIV Stigma			0.05 (−0.03 to 0.14)	0.05 (−0.03 to 0.14)
HIV status disclosure			−0.06 (−0.25 to 0.14)	−0.03 (−0.24 to 0.17)
Depression			0.25 (0.15–0.36)	0.22 (0.11–0.32)
Family and Social factors
Adults in the household				0.01 (−0.14 to 0.17)
Caregiver Communication
No communication				Ref
Communicate about HIV				1.32 (0.56–2.08)
Communicate about sex				1.09 (0.20–1.97)
Talk about bad friends				−0.36 (−1.25 to 0.54)
Talk about puberty				0.75 (−0.03 to 1.53)
Peer pressure
No peer pressure				Ref
Some peer pressure				2.51 (1.16–3.86)
A lot of peer pressure				3.37 (1.85–4.89)
Constant	9.12 (5.90–12.33)	9.15 (6.02–12.28)	4.98 (1.46–8.49)	1.62 (−2.33 to 5.56)
R2	0.0070	0.0159	0.0469	0.1143
Incremental R2		0.0089	0.0310	0.0674

Values in bold denote statistically significant results.

In model 3, we added psychological factors to model 2. The variables included in model 3 explained 4.69% of the total variance (R² = 0.0469). Specifically, adding psychological factors increased the total variance explained by 0.0310. After introducing psychological factors, adolescents working for a pay became significantly associated with more approving sexual risk-taking attitudes (β = 1.85, 95% CI: 0.37–3.33, p = .015). However, caregiver having formal employment remained significantly associated with approving sexual risk-taking attitudes (β = −0.08, 95% CI: −0.10 to −0.07, p < .001). Among the psychological factors, only depressive symptoms were associated with sexual risk-taking attitudes (β = 0.25, 95% CI: 0.15–0.36, p < .001). Specifically, for every unit increase in the depressive symptom scores, the sexual risk-taking attitudes increased by 0.25.

To fit model 4, we introduced family and social factors to model 3. Factors in model 4 explained 11.43% of the total variance (R² = 0.1143)—increasing the variance explained by 0.0674. We found that communicating with the caregiver about HIV (β = 1.32, 95% CI: 0.56–2.08, p = .001), communicating about sex (β = 1.09, 95% CI: 0.20–1.97, p = .017), and experiencing some peer pressure (β = 2.51, 95% CI: 1.16–3.86, p = .001) and a lot of peer pressure (β = 3.37, 95% CI: 1.85–4.89, p < .001) were associated with positive sexual risk-taking attitudes. In addition, caregiver having formal employment was associated with negative sexual risk-taking attitude (β = −0.08, 95% CI: −0.10 to −0.06, p < .001), while adolescents working for a pay (β = 1.78, 95% CI: 0.28–3.29, p = .022) and depressive symptoms (β = 0.22, 95% CI: 0.11–0.32, p < .001) were associated with positive sexual risk-taking attitudes.

Discussion

This study examined the incremental role of individual-level (demographic, economic, and psychological) and social factors associated with positive attitudes toward sexual risk-taking behaviors among ALHIV. Although few ALHIV in our sample reported having previously engaged in sexual intercourse, almost one-third of these had unprotected sex. Our results showed that economic and psychological factors at the microlevel and social factors at the mesolevel were associated with sexual risk-taking attitudes, which support the socioecological framework that informed the study. These findings have important programmatic implications for controlling HIV spread among adolescents.

The economic factors that were significantly associated with sexual risk-taking attitudes included caregivers having formal employment and a stable income and ALHIV working for pay. ALHIV coming from households where the primary caregiver was employed and had a stable income showed less approving sexual risk-taking attitudes, possibly because they did not need to indulge in risky sexual activities to earn a living. The findings align with literature, which consistently shows how poverty influences risky sexual behaviors among adolescents [20,21]. Lack of basic needs can drive adolescents to engage in risky sexual activities to meet their basic needs [39].These findings support the theory of welfare by assets [40], linking household financial stability to positive behavioral attitudes and orientation toward a productive future. We also found that ALHIV working for pay showed more positive attitudes toward engaging in risky sexual activities as compared to those who were not working. This association seems counterintuitive from what we would expect because adolescents who work for pay are able to earn income which reduces their pressure of engaging in sexual risk-taking behaviors in search for earning a living. However, there could be other plausible explanations why the ALHIV who work were more approving of risky sexual attitudes. For instance, some adolescents may be influenced to pick the negative influences from their workplaces. These results call for qualitative studies to further understand the influence of working for an income on sexual risk-taking in this population. This finding is consistent with results from a study in South Africa, where adolescents who were employed were more likely to have sex than those who were not employed [41]. Most likely, these adolescents were working at such an early age because of underlying economic constraints and the need to meet their basic needs. Therefore, the need to meet their basic needs could similarly drive them to engage in risky sexual activities. Another plausible explanation is that workplace exposes the ALHIV to risky sexual vices from workmates. Adolescents working for a pay only became significant when we introduced psychological factors at models 3 and 4, which suggested possible confounding relationship between economic and psychological factors in influencing sexual risk-taking among adolescents.

Among the psychological factors, we found that having more depressive symptoms was associated with more positive sexual risk-taking attitudes. This association can be explained by several mechanisms. First, depressed ALHIV usually exhibit other negative state like hopelessness, which demotivate them from adopting less risky behaviors [42]. Also, ALHIV may use risky sexual activities as a means to escape the depression [42]. Our results are consistent with findings from a study among youths in Tanzania, which found depression to influence risky sexual behaviors [43]. Hence, interventions to reduce risky sexual behaviors should also incorporate strategies to address depression because it is an important driver of risky sexual behaviors.

We found that several social factors (mesolevel) were significantly associated with sexual risk-taking attitudes among ALHIV. Overall, social factors explained the highest amount of variance in sexual risk-taking attitudes. As is the case in similar settings, in Uganda, peer pressure was associated with more approving sexual risk-taking attitudes among adolescents [20,44]. In fact, the intensity of peer pressure corresponded with strength of associations with sexual risk-taking attitudes, whereby adolescents who experienced lower levels of peer pressure showed a weaker association with sexual risk-taking attitudes than those who reported intense peer pressure. ALHIVs that experienced higher levels of peer pressure had more positive sexual risk-taking attitudes. These findings suggest a role of peer pressure in influencing sexual risk-taking among ALHIV. Programs aiming at reducing horizontal HIV transmission among adolescents should explore interventions that specifically incorporate discussions (coping strategies) about peer pressure or extend to adolescents’ peers to address factors such as peer pressure, which influence adolescent behavior.

Similar to findings from a separate analysis from the same sample, adolescent-caregiver communication about HIV and sex was associated with more positive sexual risk-taking attitudes [30]. There are several possible explanations for this relationship. As reported by Bastien et al. in their review of studies among the general population of adolescents from SSA, in many African settings, communication about sex with caregivers is considered immoral and shameful and only occurs after an undesired event such as unintended pregnancy or an unsanctioned romantic relationship [14]. Also, the cultural norms and taboos in many African settings hinder effective parent-child communication about sexuality and HIV [14,45]. Consequently, the communication is often initiated by the parent and is in the form of threats, commands, or blame [45,46]. This form of communication is bound to be ineffective in addressing negative behaviors [45]. A study by Phetla et al. done in South Africa found that adolescents were opposed to communication with their parents because of the judgmental nature of the communication [15]. Therefore, parental communication would be more effective if it is not unidirectional and provides an environment for dialogue. Studies elsewhere showed contradicting results on the relationship between adolescent-caretaker communication and sexual risk-taking behaviors. For instance, a study from Ethiopia, and another among adolescents in four African countries including Uganda, found that adolescent-caregiver discussion was protective against risky sexual behaviors [47,48]. In contrast, a study by Ismayilova et al. in Uganda found that having discussion between the adolescents and their caregivers was not associated with sexual risk-taking attitudes [49]. Also, a review by Bastien et al., which included 23 articles published between 1980 and 2011, found both positive and negative associations between adolescent-caregiver communication and sexual risk-taking behaviors [14]. Introducing social factors did not affect the significant individual-level factors already identified. Our results suggest a need for research involving interventions to overcome the cultural and contextual barriers during adolescent-caregiver communication. Additional research should seek to understand the content, tone, frequency, acceptance, and tolerability of the parent-child communication regarding HIV and sex.

Our analysis revealed that the factors included in the study explained 11% of the variance in sexual risk-taking attitudes. It is worth noting that there are additional factors, such as lack of knowledge about HIV and other STIs and substance use that have also been consistently linked to increased sexual risk-taking behaviors. While these factors were not included in our analysis, they may be important to consider in future research on this topic. Nonetheless, our findings highlight the need for further research on the drivers and pathways that influence adolescents’ intentions to engage in sexual risk-taking behaviors, particularly in low-resource communities. Adolescents are particularly vulnerable to engaging in these behaviors and understanding the factors that contribute to this vulnerability could inform the development of targeted interventions. One potential intervention that should be considered is promoting caregiver-adolescent communication as a strategy for bridging the communication gap between adolescents and their parents on issues related to sexuality and HIV. This could involve providing resources or training to caregivers to help them feel more comfortable and confident in discussing these topics with their adolescents. Additionally, exploring and understanding the sociocultural factors surrounding sexual activity among adolescents could yield valuable insights into the context in which these behaviors occur. This could inform the development of culturally sensitive interventions that are more likely to be effective in addressing the needs of adolescents in different cultural settings.

Our study had some limitations. First, less than 5% (n = 33) of the participants had engaged in sexual intercourse. This number was too small to make meaningful inferential analyses. For instance, ALHIV who had never had sexual activity might have been different from those who reported having sex. However, because of the small number of ALHIV that reported having sex, we could not compare the two groups. We used sexual risk-taking attitudes as a proxy for sexual risk behaviors. However, sexual risk-taking attitudes were found to be a reliable predictor of future sexual risk-taking behaviors [32]. Also, social desirability bias could have affected responses in the study because the data were collected using an interviewer-administered questionnaire and relied on participant self-reports. The study was conducted in a rural setting, with HIV prevalence higher than the national average [3]. HIV disrupted family structures in this region and many ALHIV are orphaned. Hence, the results should be generalized with caution, especially in urban settings and contexts with a low HIV burden. Finally, our study was among adolescents of ages 10–16 years, and only 5% of the sample reported having sex. Therefore, these results may not be generalizable to older adolescents and those who are sexually active.

Conclusion

These results add to our understanding of the factors associated with positive attitudes toward sexual risk-taking behaviors among ALHIV and may have significant ramifications in controlling HIV spread among adolescents in low-income settings. As suggested by the socioecological model, we have demonstrated that factors at both the individual and social levels are critical in influencing sexual risk-taking among ALHIV. Efforts targeting the reduction of sexual risk-taking behaviors should also address factors beyond the individual. These efforts should incorporate strategies that strengthen psychosocial support systems for the ALHIV and economic empowerment for their families.

IMPLICATIONS AND CONTRIBUTION.

These findings highlight considerable sexual risk-taking behaviors among ALHIV, including unprotected sex and more approving attitudes toward risky sexual behaviors, which undermine efforts to reduce horizontal transfer of HIV among adolescents. Programs designed to reduce HIV transmission among the adolescents need to address economic, social, and psychological factors if they are to be effective.

Data Availability

Upon submitting a reasonable request, the data we used for the analysis in this paper can be availed by the corresponding author.