Introduction

The number of Kenyan children who have lost one or both parents has grown steadily from 1.16 million in 1995 to 1.7 million in 2003 [1, 2] and 2.4 million in 2008 [3]; these numbers reflect an increasing proportion of Kenyan children younger than 15 years old who are orphaned, from 7 % in 1993 [2] to 11 % in 2007 [4]. Deaths related to HIV/AIDS among parents have contributed to this significant growth in orphanhood; of the total numbers of orphans, estimates suggest that this proportion has increased from 22 % in 1995 to 48 % in 2008 [13]. Approximately one-fifth of orphaned children in Kenya are aged 12–17 years [5].

Numerous studies in sub-Saharan Africa [2, 614] have shown that orphans may be at greater risk of acquiring HIV and other STIs than non-orphans due to a heightened likelihood of experiencing coerced sex and exploitation through transactional sex, which may be attributed to the psychosocial and economic distress resulting from loss of one or both parents [10, 14]. Most of these studies, however, have been undertaken in southern African countries, primarily South Africa [9, 10] and Zimbabwe [7, 8, 11, 14], which have very high HIV prevalence. The East African region generally has a lower prevalence of HIV and different social and cultural practices, and so the experiences of orphans may be somewhat different. In Kenya, for example, the national adult HIV prevalence rate is 7.4 %, with marked regional variations ranging from less than 1 % in North–Eastern Province to 15 % in Nyanza Province, with rates of all other provinces ranging from 4 to 6 %, except Nairobi with 9 % [4].

Premarital sexual activity is common among Kenyan adolescents aged 15–19, with about one quarter of females and nearly half of males having had sex [15]. Condom use is moderate but increasing, with about one half of sexually active males and females reporting condom use at last sex [15]. National data on transactional sex in Kenya is only available for men and is low, at less than 2 % among 15–19 year old [15].

Nyanza Province

Nyanza has some of the worst reproductive health indicators in Kenya for adolescents aged 15–19 years, including the highest HIV prevalence rate of 2.3 % (with females three and half times more likely to be infected than males), the highest proportion that have ever had sex, and low to moderate levels of condom use at last sex [15]. Transactional sex is relatively common among girls and is associated with poverty, funeral discos, alcohol and drug use, and video shows [16, 17]. The province also hosts the largest proportion of orphaned adolescents, estimated at 19 % compared to the national level of 11 % [4]. In 2004, the province had approximately 500,000 orphans, about one-third of all orphans in Kenya [18].

In-depth case studies of female adolescent orphans in Nyanza Province found that their life situations may increase exposure to HIV because their ability to adopt risk prevention measures is determined by a range of factors beyond their control [19]. However, the authors acknowledge that the case study methodology limits the generalizability of the results [19]. There is, therefore, a need for additional information on orphaned adolescents in Nyanza to build the evidence base and guide programming.

While previous research has examined demographic factors for predicting adolescent sexual risk [2025], including their orphan status [214], cultural factors may also be important. Nyanza Province is predominantly inhabited by the Luo ethnic group, whose traditional cultural beliefs and practices could influence adolescent sexual behaviors and values [15]. According to Luo traditional culture, when children (both male and female) reach adolescence, they sleep at a house separate from their parents, either within or outside the immediate family homestead. A Luo homestead often comprises several households, including houses for the homestead head’s wife (or wives in cases of polygamy) and their adult sons, both married and unmarried [26]. Girls mainly sleep at their grandparents’ house and receive moral teaching through story telling by their grandmother and older female relatives, who also chaperone them and socialize them on various issues including sexuality [2729]. Other girls sleep in the household kitchen (usually a separate structure from the main house). Boys mainly sleep at an older bachelor brother’s hut (simba), or at an uncle’s hut (traditionally equivalent to his father) and are socialized by older male family members in the evening around the fireplace (duol) where all homestead male members have their supper together. The socialization process is intended to inculcate values that help adolescents remain chaste as they grow up [27, 30, 31].

However, these traditional social practices are changing [30, 31], potentially compromising the protection previously assured, especially in terms of child nurturing. Children are increasingly sleeping in the same house as the household head, whether a parent or a caregiver; whereas monitoring their behaviour may still exist, any open discussion of sexuality between parents and their children that may help in nurturing a protective setting remains a cultural taboo [31].

Moreover, the Luo funeral cultural practices frequently last for several days, at which community members, friends, and relatives keep vigil at the home of the deceased to comfort the family, keep them company and assist in funeral arrangements. Traditionally, the night vigils were accompanied by singing dirges (mourning songs) and dancing [32]. Over time, the dirges have been complemented by popular afro-Christian songs, and/or recorded music that mourners dance to. A study of these “disco funerals” in Kisumu town (the provincial headquarters) found that such social gatherings create ample opportunities for risky sexual behaviour by adolescents, with alcohol and drugs enhancing such behaviours [17].

The purpose of this study was to establish the extent to which orphanhood, cultural and socio-demographic factors contribute to sexual risk behaviours among adolescents in Nyanza Province, western Kenya.

Methods

Study area and population

The study was conducted in August 2006 in Kisumu, Migori, Homa Bay, and Suba districts of Nyanza Province. The province lies along the shores of Lake Victoria and covers an area of 12,547 km2 with a population of more than 4.3 million [33]. Kisumu, Migori, and Homa Bay districts are mainly inhabited by the Luo, while Suba district is predominantly occupied by the Abasuba who have assimilated the Luo culture and language. Nyanza province is characterized by generalized poverty, recording Kenya’s highest poverty index of 65 % [34]. The main economic activities include subsistence farming, lake fishing and small-scale food trading.

Design and sampling

Using stratified sampling, two constituencies (political boundaries represented by a member of parliament) were randomly selected from each of the four districts. For each constituency, location (an administrative unit) was randomly chosen, within which one sub-location (the smallest administrative unit) was then chosen randomly, giving a total of eight sub-locations for the study sites.

The sample size needed to identify any meaningful differences between orphans and non-orphans in terms of behaviors and other characteristics was calculated using the estimated prevalence of orphanhood (35 %) among 14 to 17 year-olds. Based on the estimated population size of 5,800 [35], a confidence level of 95 % and a confidence interval of 3.8, the total sample size needed for the cross-sectional survey was 597. To achieve the sample size, every 11th through13th household in each sub-location (depending on the number in each) was sampled. One adolescent of each gender in the age category 14–17 years was interviewed per household; if there was more than one eligible boy or girl, only one was randomly selected.

Ethical Procedures

Ethical review and approvals for the study were obtained from the Kenyatta National Hospital Ethics and Research Committee and the Institutional Review Board of the Population Council. Parental informed consent and adolescent participant assent were obtained. There were no refusals from parents/caregivers or adolescents. Immediate on-site support in case any distress was experienced during the interviews was made available through including trained counselors of orphaned and vulnerable children (OVC) among the data collectors.

Data collection

Quantitative data were collected through a cross-sectional survey using a structured questionnaire among a representative sample of adolescent girls and boys aged 14–17 years. The questionnaire was informed by prior qualitative research that identified the main factors perceived to influence risky sexual behaviors among adolescents in the study communities. Data were collected on orphan status, psychosocial support at home, perceived parent/caregiver capacity to provide adolescent basic needs, engaging in income generating activity, parent/caregiver-adolescent discussion about sexual and reproductive health (SRH) issues, HIV/AIDS knowledge, sleeping arrangements, participation in social activities, sexual activity, engaging in transactional sex, and condom use. The terms “ever had sex” and “sexually active” are used interchangeably to mean any adolescent who has had sex. Appendix provides operational definitions of the key variables measured.

The survey was administered face-to-face in Dholuo (the local language) or in English, depending on the respondent’s language competence and preference. A team of 20 experienced interviewers, five of whom were trained counselors of OVC, were recruited and trained in the data collection procedures.

Data analysis

A total of 546 adolescents were interviewed, which represents 91 % of the required sample size. While there were no refusals, the available sample ceiling was reached. Data were analyzed using SPSS version 15.0 (IBM SPSS Statistics, September 2006). For data reduction, factor analyses using the maximum likelihood procedure were conducted, which yielded four multi-item key-constructs:

  • Parent/caregiver capacity to provide for adolescents needs

  • Psychosocial support at home

  • Discussions between parents/caregivers and adolescent on sexual and reproductive health issues

  • HIV/AIDS knowledge

Subsequently, the items that represented one dimension (one scale) were subjected to a scale-reliability analysis calculating the Cronbach’s alpha (internal consistency).

Cross-tabulations and Chi-square (χ²) tests were run for categorical background characteristics, recoded scale variables, and sexual behavior variables by orphan status for boys and girls separately to determine any association. To better understand possible underlying processes, we performed mediation analyses to establish the relation between orphan status and ever had sex [36]. Multivariate logistic regression was performed to determine the association between orphan status and selected predictor variables. Multivariate logistic regression for transactional sex and condom use at last sex were done for adolescents who reported ever engaging in sex.

Results

Socio-demographic characteristics of adolescents

Thirty-eight percent of the adolescents were orphans (12 % double orphans, 6 % maternal orphans, and 20 % paternal orphans). Table 1 shows the overall distribution of socio-demographic characteristics of study participants by gender and orphan status. Slightly over a half (54 %) of the adolescents slept in a different house from the household head, and of these, 43 % slept within the same homestead as the household head while 11 % slept outside the homestead (data not shown).

Table 1 Socio-demographic and cultural characteristics by gender and orphan status (%)

Girls were less likely than boys to sleep in a different house from their household head (χ² = 14.652 (N = 546), p ≤ 0.001), engage in income generating activity (χ² = 28.401 (N = 546), p ≤ 0.001), or attend discos/other night social activities (χ² = 32.791 (N = 546), p ≤ 0.001). Girls reported less psychosocial support at home (χ² = 13.183 (N = 546), p ≤ 0.001), and had less knowledge about HIV/AIDS (χ² = 6.055 (N = 209), p = 0.048).

Orphans were more likely than non-orphans to report having engaged in income generating activity (χ² = 7.294 (N = 546), p ≤ 0.008) and attending disco/other night social activities (χ² = 4.866 (N = 546), p ≤ 0.034). Orphans were also less likely than non-orphans to report high caregiver capacity to provide for their basic needs (χ² = 23.567 (N = 546), p ≤ 0.001) or to discuss sexual and reproductive health issues with their parents/caregivers (χ² = 8.427 (N = 546), p ≤ 0.015). Overall, most adolescents reported moderate-to-high levels of psychosocial support at home, although many had never discussed pregnancy, STI or HIV infection prevention with their parents/caregivers. More than a half (56 %) of the adolescents reported moderate HIV/AIDS knowledge (3–5 correct responses).

Sexual Risk Behaviors

Table 1 also shows that 44 % of all adolescents reported ever having had sex. Of these, 42 % had engaged in transactional sex and 25 % reported condom use at last sex. Significantly more boys than girls (50 vs. 37 %) had ever had sex (χ² = 8.530 (N = 48), p ≤ 0.004), while among those who had ever had sex, girls were more likely than boys to have engaged in transactional sex (52 vs. 36 %) (χ² = 5.154, (N = 209), p ≤ 0.031). There were no significant gender differences in condom use at last sex. Of those who had ever had sex, 31 % of boys and1 % of girls reported ever giving money, a favor or gift (data not shown in Table 1).

Significantly more orphaned adolescents reported ever having had sex (50 %) compared to non-orphans (40 %) (χ² = 4.073 (N = 483), p ≤ 0.047). To further explore this relationship, mediation analyses were performed using logistic regression according to the four-step approach suggested by Barron and Kenny [36]. The significant relationship between orphan status and ‘ever had sex’ (Beta 1.46; p = 0.04) (step 1) was substantially reduced after ‘perceived caregiver capacity to provide for the adolescent’s needs’ was added as a mediator in the equation (Beta 1.24; p = 0.27) (step 3), suggesting a partial mediation. The relationship between orphan status and ‘ever had sex’ was also reduced after ‘attending night activities’ was included as a mediator (Beta 1.34; p = 0.14) (step 3), also suggesting partial mediation. The combined indirect effects of these two mediating variables resulted in a nearly zero effect of orphan status on ‘ever had sex’ (Beta 1.09; p = 0.67) (steps 3 and 4) and both variables were significantly related to orphan status (Beta 1.10; p = ≤ 0.001 and Beta .24; p ≤ 0.001 respectively) (step 2).

These results suggest that participation by orphans in night activities and the perceptions of adolescents that their parents/caregivers are unable to provide for their basic needs are more closely associated with the likelihood of them ever having engaged in sex than their orphan status. Sleeping place, HIV knowledge, residence (town vs. rural), psychosocial support and discussion of reproductive health issues with a parent/caregiver were not associated with this relationship.

Multivariate analyses of predictors of sexual activity, transactional sex and condom use at last sex

Multivariate logistic regression analyses for predictors of ever having had sex, having engaged in transactional sex and condom use at last sex are shown in Table 2. No association was found between orphanhood and engaging in sex, transactional sex or condom use at last sex. Older adolescents were more likely to be sexually active (OR = 1.23, CI = 1.02–1.48). Adolescents living in urban areas were less likely to have had sex than their rural counterparts (OR = 0.54, CI = 0.34–0.87), 29 versus 56 % respectively. Sleeping in a different house from the household head was a strong predictor of ever having had sex (OR = 1.75, CI = 1.09–2.79). Also, adolescents who attended disco or other night social activities were much more likely than those who did not attend such functions to have ever had sex (OR = 3.7, CI = 2.38–5.63). Additionally, adolescents who perceived their caregivers to have the ability to provide for their basic needs were less likely to have engaged in sex (OR = 0.93, CI = 0.88–0.97). These results are in line with the mediation analysis which also showed that orphanhood is not directly associated with having engaged in sex. Engaging in income generating activities, family psychosocial support, parent/caregiver-adolescent discussion about sexual and reproductive health issues, and HIV/AIDS knowledge were not associated with ever had sex.

Table 2 Multivariate logistic regression of socio-demographic and cultural characteristics on ever had sex, engaged in transactional sex and condom use at last sex

For adolescents who ever had sex, girls were three times more likely than boys to be involved in transactional sex (OR = 3.23, CI = 1.58–6.58) (Table 2). Transactional sex was also more likely associated with sleeping outside the parents’/caregivers’ house (OR = 2.16, CI = 1.04–4.52) and with attending disco and other night activities (OR = 2.03, CI = 1.00–4.13). When age of the adolescent is taken into account, however, the relation between sleeping arrangement and transactional sex only holds for the older (>16 years) adolescents (χ² = 7.480, (N = 113), p = 0.005), with older adolescents sleeping in a different house from the household head (54 %) being more likely to have had transactional sex than older adolescents who slept in their parents’/caregivers’ house (26 %). The adolescents’ age did not affect the relation between attending discos and other night activities and transactional sex. No association was found between transactional sex and orphanhood, urban versus rural living, engaging in income generating activities, psychosocial support at home, perceived ability of parent/caregiver to provide for adolescent basic needs, and parent/caregiver-child discussion about pregnancy/STD/HIV.

Condom use at last sex was more likely among older adolescents (OR = 1.52, CI = 1.10–2.09) and among adolescents who discussed sexual and reproductive health issues with their parents/caregivers (OR = 1.52, CI = 1.14–2.02). Orphanhood and other predictors were not associated with condom use at last sex.

Discussion

Contrary to studies conducted with adolescents in other African countries [2, 4, 615], orphanhood does not appear to be significantly associated with sexual activity, transactional sex, or condom use at last sex among adolescents aged 14–17 years in this population. The mediating effects of attending night social activities, including funeral gatherings, and a perceived inability of the parent or caregiver to provide their basic needs helps to explain the lack of a direct association between orphanhood and sexual activity. These findings suggest that there are predictors other than being orphaned that are more likely to put young people at risk of engaging in sex. Moreover, adolescents living in urban areas (Kisumu and Homa bay towns) were less likely to have had sex compared to adolescents from rural areas, which is contrary to the general perceptions in Kenya that urban life exposes youth to greater opportunities for sexual activity.

Our study also showed that sleeping in a different house from the household head was associated with a greater likelihood of ever having sex, for both adolescent boys and girls, irrespective of their orphan status. The majority of adolescents who did not sleep in the same house as the household head slept in a house located within the same homestead, including their grandparents or brothers (often married) and so their behavior may not be keenly monitored as those who slept in the same house as the household head.

Traditionally, chaperoning and counseling of adolescents towards initiation into adulthood were integrated into their sleeping arrangements and were conducted by a grandmother or aunt for girls and a grandfather, uncle or other adult male relative for boys. However, the nature of these practices may have changed as children and these “socialization agents” no longer spend much time together [31], and so children now have other sources of information, which may not be protective.

The strong association between attending disco/night social activities (including night funeral discos) and having had sex concurs with a previous qualitative study among adolescents in Nyanza province [18]. Traditionally, selected youth, chaperoned by adults, were allowed to be present so that they could run errands such as keeping the fires burning and serving hot tea [32]. The chaperoning roles previously played by adults are now largely lost, as adults often do not stay throughout the night, leaving the youth alone to enjoy the company and music, providing opportunities for sexual activity. Peer pressure during the night can enhance risk taking [17].

The perceived incapacity of parents/caregivers to provide for the adolescents’ basic needs could be explained by caregivers being ignorant of the things that adolescents, and in particular girls, want or need; or they may lack the financial resources to provide for these. A better understanding of the association between the adolescents’ perceptions of their parents’/caregivers’ ability to provide for their basic needs and their likelihood of engaging in sex may help inform parents and caregivers to better set priorities in addressing adolescents’ needs and discussing these priorities with their children. Parallel strategies are also needed to equip parents and caregivers who are too poor to provide these basic needs with the resources to do so, possibly through appropriate income generating activities. Additionally, there is need to strengthen and expand existing social protection systems. An example is the cash transfer program for OVC (CT-OVC) that provides regular and predictable cash transfers to extremely poor families living with OVC in order to encourage fostering and retention of OVC within their families and communities, and to promote their human capital development. These include school enrolment and attendance, household nutrition and food security, improved health, and increased civil registration of children and caregivers [37].

Although girls were less likely to have had sex, those who were sexually active were more likely than boys to have engaged in transactional sex. How negotiation about transactional sex happens and what motives play a role in these processes is not clear from these data. A better understanding of such processes may provide suggestions for how, for example, safe sex behaviors like condom use, could be integrated in teaching adolescents safe sex negotiation.

Limitations

Self-reports of sexual and other behaviors may not always be valid, because girls sometimes under-report while boys tend to exaggerate their sexual activity [38, 39]. The data for this study were collected through surveys administered in face-to-face interviews and thus subject to interviewer-respondent interaction effects on answers to sensitive questions. Participants may have reported what they wanted the interviewer to hear rather than what they actually thought or did.

Secondly, the sample size was not adequate to allow a full analysis by sub-categories, for example, orphan status in terms of maternal, paternal or double orphans. One study with a larger sample size in Zimbabwe found significantly greater prevalence of HSV-2 among both maternal and paternal orphans’ sub-groups but not for double orphans compared with non-orphans [7]. Other studies with larger sample sizes in Zimbabwe [8, 11] and South Africa [10] found that maternal orphans were more likely to have engaged in risky sexual behaviors, such as unprotected sex, than paternal or double orphans or non-orphans. A larger sample size would have allowed such analysis to identify what categories of orphans are most vulnerable.

The study was conducted in an area of generalized poverty as well as a generalized HIV epidemic. The results, therefore, cannot be generalized to the Kenyan population as a whole. However, the results do provide useful insights and suggest possible interventions to address sexual and reproductive health risks for adolescents, including orphans, living in this situation.

Further research with a larger sample is recommended that would allow for analysis by orphan status. A qualitative study that explores factors around sleeping practices for adolescents would inform ways of reinforcing the traditional protections that were incorporated into these practices. Further inquiry into how adolescents initiate and negotiate having sex, including transactional issues, may help to understand how adolescents’ skills could be enhanced to include safe sex behavior in such negotiations. Based on our findings we also recommend research focusing on the development and evaluation of community-based caregivers’ capacity strengthening, including enhancement of their self-efficacy and skills in discussing sexual and reproductive health issues and safe sex practices with their adolescent children in a changing and modernizing society.

Conclusion

The findings suggest that, in this population, orphan status is not associated with increased sexual risks among adolescents. Factors other than orphan status increase adolescents’ sexual risks and apply to adolescents generally. Key among these are unsupervised sleeping arrangements, night vigils/disco during funerals, and perceived parental incapacity to provide for basic grooming needs and clothing. Interventions should address these risk factors especially in so far as they may contribute to early sexual debut and transactional sex. Sexual and reproductive health programs should target all adolescents, both orphans and non-orphans, and together with adolescents and parents, identify appropriate acceptable community structures for implementation through capacity building of community-owned resource persons. Such programs could provide counseling and information to help youth to make informed decisions on sexual behavior and could also include elements to more effectively enhance parents’/caregiver’s capacity to discuss SRH with their children, strengthen their capacity to provide for adolescents’ basic needs and initiate income generating activities for adolescents. Such initiatives should target all families and adolescents in rural as well as urban areas, regardless of orphan status. Additionally, programs should build adolescents’ life skills and increase their access to reproductive health information and services through provision of youth friendly services and centres for all adolescents.

While traditional practices cannot be changed overnight, the transformation of Luo cultural practices around sleeping arrangements and funeral night vigil indicate that culture is not static. It is imperative to reassess the place of these traditions in current social space, and the best safety enhancing strategies for their continued practice. It may be useful to introduce programs that challenge communities to question the value, if not of culture, at least of aspects of culture as they are currently practiced. The aim would be to aid reflection on aspects of practices that have perhaps lost their intended meaning and structure. If the practices are deemed to still have a place in society, then ways of revamping the original intended meanings and practicable supportive structures should be explored with a view to minimizing negative consequences associated with their current practice. Policy makers and programmers could use this information to advance efforts to find best practices for resolving the conflict between transformations of certain aspects of culture and public health interventions such as those advanced for HIV/AIDS.