Orphan Status, HIV Risk Behavior, and Mental Health Among Adolescents in Rural Kenya

Abstract

Objective To examine orphan status, mental health, social support, and HIV risk among adolescents in rural Kenya. Methods Randomly selected adolescents aged 10–18 years completed surveys assessing sexual activity, sex-related beliefs and self-efficacy, mental health, social support, caregiver–child communication, time since parental death, and economic resources. Analysis of covariance and regression analyses compared orphans and nonorphans; orphan status was tested as a moderator between well-being and HIV risk. Results Orphans reported poorer mental health, less social support, and fewer material resources. They did not differ from nonorphans on HIV risk indicators. Longer time since parental death was associated with poorer outcomes. In moderator analyses, emotional problems and poorer caregiver–youth communication were more strongly associated with lower sex-related self-efficacy for orphans. Conclusions Orphans are at higher risk for psychosocial problems. These problems may affect orphans’ self-efficacy for safer sex practices more than nonorphans. Decreased HIV risk could be one benefit of psychosocial interventions for orphans.

Introduction

Kenya has 2.4 million orphans, roughly half of which have lost one or both parents to HIV/AIDS (National AIDS Control Council, 2008). Compounding the challenges of losing a parent, orphans may be at increased risk for HIV. At least four studies in southern Africa have documented higher rates of HIV among orphans (Birdthistle et al., 2008; Gregson et al., 2005; Nyirenda, McGrath, & Newell, 2010; Operario, Pettifor, Cluver, MacPhail, & Rees, 2007). Studies suggest that increased rates of sexual risk behavior may help to explain this disparity. Orphans have been found to have a higher likelihood of being sexually active, initiate sex at an earlier age, and have unprotected sex and multiple partners (Birdthistle et al., 2008, Gregson et al., 2005; Juma, Askew, & Ferguson, 2007; Operario et al., 2007;Nyirenda et al., 2010).

Orphans’ risk for contracting HIV is particularly important in high prevalence areas where it is common to lose parents to AIDS-related deaths. Nyanza Province is one such area with an HIV prevalence of 14.9%, twice the national average (Gelmon, Kenya, Oguya, Cheluget, & Haile, 2009). This study was conducted in Nyanza in Muhuru Bay, a district along the eastern shore of Lake Victoria where fishing is the main source of income. Because of high mobility and transactional sex, fishing communities are particularly vulnerable, with estimated HIV prevalence between 25–30% (Gelmon et al., 2009). An estimated 35% of all youths in this province are orphans (Juma et al., 2007). Given that these children are in a high-risk environment and many have lost a parent to HIV/AIDS, it is important to understand their psychosocial and economic challenges relative to other children. It is also important to understand how these challenges relate to their risk for contracting HIV themselves.

Pathways leading to increased prevalence of HIV and risk behavior among orphans remain unclear. In the general population of children, primarily in the Unites States, researchers have documented that poor psychosocial well-being likely increases risk behavior (Malow, Rosenberg, Donenberg, & Devieux, 2006), whereas positive child–caregiver relationships are protective (Perrino, Gonzalez-Soldevilla, Pantin, & Szapocznik, 2000). A previous study on psychosocial correlates of HIV risk behaviors among adolescents in this sample (Puffer et al., 2011) also identified emotional symptoms as a risk factor. It is possible that such factors are compounded for orphans. Poor mental health, lack of or negative caregiver social support, and inadequate economic resources have been cited as possible factors that may increase orphans’ likelihood of engaging in behaviors associated with HIV risk (Nyirenda et al., 2010; Operario et al., 2007).

If orphans are at increased risk for HIV and if psychosocial and economic challenges contribute to this elevated risk, it becomes critical to target orphans for HIV prevention efforts that aim to mitigate these factors. On the other hand, if orphans are embedded in an environment where HIV risk is elevated for all youth, then addressing youth as a whole could avoid unnecessary isolation and stigmatization that might accompany a focus on orphans (UNICEF, 2009). This study aims to add to the public health literature informing the role that orphan status should play in identifying subpopulations of youth in Sub-Saharan Africa (SSA) for HIV prevention and psychosocial intervention.

Mental Health

There is a clear pattern of higher internalizing problems among orphans in SSA including more symptoms of depression, anxiety, and posttraumatic stress disorder (Cluver & Gardner, 2007). Data on externalizing problems are mixed, with higher conduct problems among orphans in studies in Uganda and Ghana (Atwine, Cantor-Graae, & Bajunirwe, 2005; Doku, 2009), but not in South Africa (Cluver, Gardner, & Operario, 2007). In the United States, both internalizing and externalizing problems have been associated with high-risk sexual behavior. (Brown et al., 2006; Tubman, Andres, Wagner, & Artigues, 2003). Mechanisms may include compromised decision making, maladaptive coping, and low self-efficacy (Brown, Houck, Grossman, Lescano, & Frenkel, 2008). More research is needed on these associations in SSA and among orphans specifically.

Caregiver and Social Support Systems

Loss of a parent disrupts interpersonal relationships, requiring children to build or shift relationships with new caregivers. In some cases, the death can also lead to decreased social support overall, particularly if children are required to move. Orphans may be more likely to engage in early sexual activity to facilitate social connections. Further, they may lack the communication with caregivers that has been identified as a potential protective factor (Perrino et al., 2000). Qualitative evidence specifically from Nyanza indicates that community leaders, caregivers, and youth perceive that lack of caregiver supervision and emotional support are among the main reasons that orphans engage in risky behavior (Juma et al., 2007). Orphans also may not receive the adequate caregiver monitoring that deters risky sexual behavior (DiClemente et al., 2001). In fishing communities especially, female orphans with less supervision may be more likely to engage in transactional sex (Juma et al., 2007). In general, the number of orphans may overwhelm existing support systems, decreasing the available caregiver and family influences that may prevent risky sexual behavior.

Economic Resources

According to the Demographic and Health Survey, households in Kenya that care for orphans do not have systematically lower household incomes (Bicego, Rutstein, & Johnston, 2003). However, orphans may experience socioeconomic disadvantages because of uneven distribution of wealth within households. Studies in Nyanza have documented that orphans are more likely than nonorphans to lack basic necessities (Juma et al., 2007; Nyambedha, Wandibba, & Aagaard-Hansen, 2003). In Luo communities (the largest ethnic group in Nyanza), polygyny is common, and tensions between cowives regarding distribution of household resources may contribute to these disadvantages for orphans (Nyambedha et al., 2003). Socioeconomic disparities may also directly affect HIV risk, as orphans may engage in transactional sex to meet their needs (Juma et al., 2007).

Parental Death

Previous literature documents that some orphans experience an acute period of both behavioral and emotional symptomatology, including dysphoria and anxiety, which typically occurs and then improves during the first 1–2 years following parental death (Dowdney, 2000). Some limited evidence suggests that a proportion of children will experience delayed onset of psychological symptoms after 2 years (Worden & Silverman, 1996). However, since most studies in developing countries included only a 1-year follow-up, little evidence is available on the difficulties that children experience at later time points.

A previous study on psychosocial correlates of HIV risk behaviors of adolescents in this sample (Puffer et al., 2011) demonstrated relationships between sexual risk behavior, mental health, and social support variables among youth. In this study, we examine indicators of HIV risk, mental health, social support, and economic resources among orphans and nonorphans in Muhuru Bay, Nyanza Province, Kenya. Our aims are to describe differences between orphans and nonorphans and to examine whether psychosocial difficulties affect orphans and nonorphans differently in terms of HIV risk.

This study is among the first to use a single sample to examine both sexual activity and numerous indicators of orphan well-being across the domains of mental health, caregiver social support and communication, and economic resources. We tested two hypotheses. First, we expected orphans to exhibit higher rates of sexual activity, beliefs more accepting of sexual risk behavior, lower sex-related self-efficacy, poorer mental health, less social support from caregivers, and fewer economic resources, with double orphans exhibiting the most difficulties. We expected that a shorter time since most recent parental death would be associated with more difficulties. Second, we hypothesized that the associations between mental health, social support from caregivers, and socioeconomic variables with HIV risk would be moderated by orphan status. We expected that psychosocial challenges would be more strongly associated with indicators of HIV risk among orphans compared with their nonorphan peers.

Methods

Participants

Adolescents were selected randomly from students enrolled in standards five through eight in the 14 primary schools in Muhuru Bay. Primary school in Kenya includes eight standards (grade levels) roughly equivalent to the grade levels in the United States educational system. All children enrolled in these standards and between the ages of 10 and 18 years were eligible to participate. A school-based sampling frame was used for logistical reasons and because previous studies have found that most youth (97%) in Nyanza Province are enrolled in school (Juma et al., 2007).

In total, 1,847 students were on school rosters, and sampling was stratified by school size. Of 353 students, 325 completed the survey (92.1%). The majority of the remaining students could not be located. Participants included 158 males and 167 female students aged 10 to 18 years (M = 14.0, standard deviation [SD] = 1.7). Most belonged to the Luo tribe (55%), and others were Suba (43%) or Luo/Suba (1%). The Luo and Suba tribes are similar culturally, and polygyny is common in both tribes. Median monthly per capita household income was 188.24 Kenyan Shillings (KSH; $1.87), with a wide range from 0 KSH to 5000 KSH ($62.58); the vast majority of responses were less than 2000 KSH.

Of 325 participants, 150 (45%) identified as orphans, with 104 single orphans (mean age = 13.96 years, SD = 1.65; 50% male) and 46 double orphans (mean age = 14.28 years, SD = 1.47; 46% male). Most single orphans had lost fathers (70%) and lived with the surviving parent (64%). Most double orphans were cared for by relatives, including older female relatives (46%), older male relatives (22%), brothers (13%) or sisters (9%), and grandmothers (4%) or grandfathers (2%); few (4%) listed nonrelatives.

Procedures

Prior to beginning activities, caregiver consent and youth assent were obtained. Participants completed a 90-minute structured interview with trained research assistants. All measures were translated into Dholuo, the local language, and back translated. Each survey item was reviewed by bilingual members of the research team and reworded to improve clarity when needed. Small adaptations were made to items to make them relevant to the context (e.g., changing examples of responses given as part of the question). Interviews were administered in private areas, and bar soaps were given as a token of appreciation. Ethics approval for the study was obtained from the Institutional Review Boards at Duke University and the Kenya Medical Research Institute.

Measures

The following are constructs and measures used in this study. All Cronbach’s α values refer to data derived from this study.

Orphan Status

Participants self-reported whether their biological father and mother were alive. Participants were categorized as nonorphan (both parents alive), single orphan (mother or father is alive), or double orphan (neither parent alive).

Time Since Parental Death

Orphans reported amount of time since each parent’s death on a Likert scale; <1 year, 1 to 3 years, 4 to 6 years, or >6 years. For double orphans, the time since the most recent parental death was used in analyses.

Sexual Activity, Self-Efficacy, and Beliefs About Sexual Risk Behavior

One item assessed history of sexual activity (yes/no), including vaginal, anal, or oral sex. Sex-related self-efficacy for condom use and sex refusal were measured with five items adapted from those used previously in SSA (Sayles et al., 2006; α = .64). Sex-related beliefs were assessed with 16 items about acceptability of risk behaviors/condom use (α = .68).

Caregiver Relationship and Social Support Factors

Participants were asked about the adult(s) most involved in their daily care. Caregiver monitoring was assessed with seven items adapted from an existing measure (Baptiste et al., 2006; α = .53). Caregiver social support was measured with eight items from the Parental Social Support for Adolescents Scale (Aneshensel & Sucoff, 1996). Frequency of communication about sex and HIV was assessed using five items from the Parent Adolescent Communication Scale (Sales et al., 2008) and two additional items about puberty and circumcision (α = .81). Sixteen items from the Parent/Adolescent Communication–Jaccard measure (Jaccard, Dittus, & Gordon, 2000) assessed barriers in communication about sex/HIV (α = .85). The Social Support for Adolescents Scale (Seidman et al., 1995) assessed social support provided by 14 types of people (e.g., parent and church members; α = .95).

Mental Health

The Strengths and Difficulties Questionnaire (SDQ) assessed emotional and conduct problems (α = .59, entire scale) (Goodman, Meltzer, & Bailey, 1998), and the Children’s Depression Inventory (Kovacs, 1985; α = .66) was used to assess depression symptoms. Traumatic experiences and stress were assessed with items from the Things I Have Seen and Heard Child Self-Report (Richters & Martinez, 1992) and the University of California at Los Angeles Post-traumatic Stress Disorder (PTSD) Reaction Index for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (Steinberg, Brymer, Decker, & Pynoos, 2004); intrusion, arousal, and avoidance symptoms were analyzed separately. The Rosenberg Self-Esteem Scale (Rosenberg, 1989) assessed self-esteem (α = .69).

Economic Resources

Participants were asked: (a) if they owned their own pair of shoes, (b) if they owned at least two sets of clothes, and (c) if they had ever lacked school-related fees (individual-level wealth indicators). Caregivers reported average monthly income; per capita household income was calculated by dividing by number of individuals in the household (household-level wealth indicator).

Data Analysis

We computed participant characteristics by orphan status using means and proportions. For the first hypothesis, we conducted analysis of covariance (ANCOVA) and multinomial regression to determine whether orphan status was associated with: (a) HIV-related risk behaviors and psychosocial factors (i.e., beliefs about risk behavior and sex-related self-efficacy), (b) mental health, (c) caregiver relationship and social support, and (d) economic resources. For variables associated significantly with orphan status, differences between orphan status groups were disaggregated using post hoc contrasts (i.e., nonorphan vs. single orphan, nonorphan vs. double orphan, and single vs. double orphan). ANCOVA and logistic regression were used to determine whether time since most recent parental death was associated with HIV risk, psychosocial, and economic variables.

For the second hypothesis, we used hierarchical regression to test whether orphan status moderated relationships between the predictor variables (psychosocial, social support, and economic) and HIV risk factors. The general equation was: where Y is the HIV-related outcome variable, X is the psychosocial/economic variable, Z is orphan status, and XZ is the interaction term (psychosocial variable × orphan status). Separate analyses were conducted for each psychosocial/economic variable. The psychosocial/economic variable was entered in Step 1, orphan status in Step 2, and the interaction term in Step 3. A significant change in variance explained by the model (R²) after Step 3 indicated significant moderation. Finally, when results indicated significant moderation by orphan status, separate ANCOVA and logistic regression models were used to examine differences between the orphan status groups (nonorphans, single orphans, or double orphans) in regards to the associations between psychosocial and HIV risk. Given previous study results showing significant psychosocial differences across age and gender, all analyses controlled for these variables. All analyses were conducted in SPPS 17.0.

Results are reported as significant at an α level of .05. However, it should be noted that after a Bonferroni adjustment for multiple statistical tests, the more conservative α level is .01; p values are presented for all results to enable interpretation at either the .05 or .01 level of significance. With a sample of 325 and an α level of .05 or .01, statistical power was adequate (>.80) to detect small effect sizes.

Results

Hypothesis 1

Table I presents results of ANCOVA analyses of associations between orphan status and HIV-related, mental health, and social support variables. Orphan status was not associated with sexual activity (odds ratio [OR] = 1.29, 95% confidence interval [CI] [0.75, 2.22]), sex-related self-efficacy, or beliefs about sexual risk behavior. For mental health, orphan status was associated significantly with emotional problems, depression symptoms, exposure to traumatic events, and intrusion and arousal symptoms of PTSD. Orphan status also was significantly associated with less social support (overall and specifically from caregivers), but not with levels of caregiver monitoring or communication with caregivers about sex/HIV. For economic resources, orphan status was associated significantly with lower access to shoes (OR = 0.32, 95% CI [0.19, 0.53]), clothing (OR = 0.43, 95% CI [0.20, 0.95]), and school fees (OR = 1.91, 95% CI [1.16, 3.15]), but not with per capita monthly household income.

Table I.

ANCOVA (Analysis of Covariance) Results of Differences in HIV-related, Mental Health, Social Support, and Economic Variables Across Orphan Status Categories (Nonorphans, Single Orphans, and Double Orphans)

Continuous outcome	F (df)	Partial η2	p
HIV risk
Risky sex beliefs	1.36 (2, 317)	.00	.95
Sex-related self-efficacy	.47 (2, 319)	.00	.63
Mental health
SDQ: emotion	7.12 (2, 320)	.04	.00**
SDQ: conduct	1.20 (2, 320)	.01	.30
Depression symptoms	7.98 (2, 319)	.05	.00**
Number of traumatic events	3.32 (2, 320)	.02	.04*
PTSD Index: total	2.66 (2, 284)	.02	.07
PTSD: intrusion	7.80 (2, 320)	.05	.00**
PTSD: avoidance	.35 (2, 318)	.01	.35
PTSD: arousal	4.51 (2, 319)	.03	.01*
Self-esteem	2.27 (2, 320)	.01	.11
Social support
General social support	12.13 (2, 319)	.11	.00**
Caregiver social support	3.67 (2, 319)	.02	.03*
Caregiver monitoring	.28 (2, 320)	.00	.75
Communication: frequency	.10 (2, 319)	.00	.90
Communication: barriers	1.86 (3, 319)	.01	.16
Economic resources
Monthly per capita income	2.38 (2, 314)	.02	.09

Note. Analyses controlled for age and gender. SDQ = Strengths and Difficulties Questionnaire; PTSD = posttraumatic stress disorder. Orphan status was coded as a categorical variable with three levels.

*p < .05, **p < .01.

Table II shows descriptive data across orphan groups and results of follow-up pairwise contrasts. These follow-up comparisons showed that both single and double orphans reported more emotional problems and depression symptoms, less social support, and less access to basic needs than nonorphans. However, single orphans did not differ significantly from nonorphans with respect to traumatic experiences, PTSD symptoms, or reported social support from caregivers. In contrast, double orphans differed significantly from nonorphans on all of these factors. Further, compared with single orphans, double orphans reported significantly more intrusive and arousal PTSD symptoms and significantly less overall social support.

Table II.

Descriptive Data on Orphan Status Groups and Pairwise Contrasts on HIV-Related, Mental Health, Support, and Economic Variables

Continuous outcome	Nonorphana (n = 175)	Single orphanb (n = 104)	Double orphanc (n = 46)
	M (SD)	M (SD)	M (SD)
Demographic variables
Age	13.88 (1.58)	13.96 (1.65)	14.28 (1.47)
HIV risk
Risky sex beliefs	28.50 (4.78)	28.55 (4.58)	28.24 (4.16)
Sex-related self-efficacy	19.26 (4.97)	18.91 (4.69)	18.51 (6.29)
Mental health
SDQ: emotion	3.74 (2.21)b,c	4.65 (2.34)a	4.89 (2.65)a
SDQ: conduct	0.92 (1.12)	0.80 (1.14)	0.67 (.920)
Depression symptoms	6.70 (3.61)b,c	8.22 (4.74)a	9.26 (5.17)a
Number of traumatic events	3.62 (2.85)c	4.20 (2.95)	4.78 (3.01)a
PTSD Index: total score	13.20 (5.90)	13.25 (5.59)	15.52 (5.44)
PTSD: intrusion	3.16 (2.17)c	3.28 (1.98)c	4.54 (2.39)a,b
PTSD: avoidance	4.83 (2.65)	5.21 (2.64)	5.35 (2.33)
PTSD: arousal	4.44 (2.63)c	4.53 (2.30)c	5.72 (2.54)a,b
Self-esteem	21.37 (3.43)	20.49 (3.93)	20.43 (3.88)
Social support
General social support	40.86 (13.10)b,c	35.14 (10.83)a,c	29.33 (9.84)a,b
Caregiver social support	27.61 (3.30)c	27.37 (3.75)c	25.87 (5.13)a,b
Caregiver monitoring	24.59 (2.65)	24.56 (2.63)	24.85 (3.33)
Communication: frequency	6.43 (5.64)	6.12 (5.80)	6.63 (5.78)
Communication: barriers	36.95 (9.77)	39.23 (10.12)	39.04 (13.14)
Economic resources
Monthly per capita income	433.80 (590.47)	290.33 (397.32)	468.14 (861.39)
Dichotomous outcome	n (%)	n (%)	n (%)
Demographic variables
Male	85 (48.60)	52 (50.00)	21 (45.70)
HIV risk
History of sexual activity	65 (37.10)	45 (43.30)	21 (45.70)
(yes/no)
Economic resources
Owns shoes	105 (40.00)b,c	34 (32.70)a	10 (21.70)a
Owns two sets of clothes	162 (92.60)b,c	88 (84.60)a	11 (23.90)a
Has ever lacked school fees	85 (48.60)b,c	67 (64.40)a	25 (54.30)a

Notes. Analyses controlled for age and gender. The indicators correspond to the three orphan status groups; they denote statistically significant (p < .05) differences in the outcome variable between orphan status groups. Those in bold and underlined denote significance at p < .01. SDQ = Strengths and Difficulties Questionnaire; PTSD = posttraumatic stress disorder; SD = standard deviation.

^aNonorphan status group.

^bSingle orphan status group.

^cDouble orphan status group.

Among orphans, half (49%) reported >6 years elapsed since parental death. Approximately one fourth (27%) reported parental death in the previous 1–3 years, 19% in the previous 4–6 years, and 5% within the past year. Given the literature suggesting an acute grief period in the first years following a parental death, time since parental death was dichotomized for analyses (0–3 years = 0, >3 years = 1). Tables III and IV show associations of time since parental death and outcome variables. Longer time since parental death (i.e., >3 years) was associated significantly with higher conduct problems, more symptoms of PTSD overall, more avoidant symptoms of traumatic stress, and more barriers to communication with caregivers about sex/HIV.

Table III.

ANCOVA (Analysis of Covariance) Results of Differences on Psychosocial HIV-Related, Mental Health, Social Support, and Economic Variables Related to Time Elapsed Since Most Recent Parent Death (N = 150, Number of Single Orphans = 104, and Number of Double Orphans = 46)

Continuous outcome	Time since parental death		F (df)	Partial η2	p
	0–3 years (n = 48)	> 3 years (n = 102)
	M (SD)	M (SD)
HIV risk
Risky sex beliefs	28.64 (4.33)	28.38 (4.51)	0.05 (1, 144)	.00	.82
Sex-related self-efficacy	19.51 (5.00)	18.46 (5.28)	0.97 (1, 145)	.01	.33
Mental health
SDQ: emotion	4.35 (2.28)	4.90 (2.50)	2.23 (1, 146)	.02	.14
SDQ: conduct	0.40 (0.74)	0.93 (1.16)	10.21 (1, 146)	.07	.00**
Depression symptoms	8.19 (3.91)	8.71 (5.29)	0.49 (1, 146)	.00	.49
Number traumatic events	4.29 (2.78)	4.42 (3.07)	0.19 (1, 146)	.00	.66
PTSD Index: total score	12.60 (5.62)	14.67 (5.23)	5.18 (1, 132)	.04	.03*
PTSD: intrusion	3.77 (2.42)	3.62 (2.19)	0.02 (1, 146)	.00	.89
PTSD: avoidance	4.19 (2.12)	5.76 (2.58)	12.79 (1, 145)	.08	.00**
PTSD: arousal	4.67 (2.40)	5.00 (2.45)	0.75 (1, 146)	.01	.39
Self-esteem	20.98 (3.49)	20.24 (4.07)	0.89 (1, 146)	.01	.35
Social support
General social support	33.71 (12.78)	33.20 (9.86)	0.01 (1, 146)	.00	.93
Caregiver social support	26.75 (3.93)	26.99 (4.42)	0.24 (1, 146)	.00	.62
Caregiver monitoring	24.40 (3.00)	24.76 (2.79)	0.27 (1, 146)	.00	.61
Communication: frequency	5.58 (5.21)	6.61 (6.02)	0.79 (1, 146)	.01	.38
Communication: barriers	35.73 (7.23)	40.81 (12.21)	8.97 (1, 145)	.06	.00**
Economic resources
Monthly per capita income	283.40 (327.36)	375.31 (673.96)	1.11 (1, 141)	.01	.29

Notes. Analyses controlled for age and gender. SDQ = Strengths and Difficulties Questionnaire; PTSD = posttraumatic stress disorder; SD = standard deviation.

*p < .05, **p < .01.

Table IV.

ANCOVA (Analysis of Covariance) Results for Associations Between Time Elapsed Since Most Recent Parental Death and HIV-Related, Mental Health, Social Support, and Economic Variables (N = 150, Number of Single Orphans = 104, Number of Double Orphans = 46)

Dichotomous outcome	Time since parental death		B (SE)	OR	95% CI	p
	0–3 years (n = 48)	>3 years n = 102
	n (%)	n (%)
Demographic variables
Male	28 (58.30)	45 (44.10)	.57 (.35)	1.77	[.89, 3.55]	.11
HIV risk
History of sexual activity (yes/no)	22 (45.80)	44 (43.10)	.05 (.39)	1.05	[.49, 2.24]	.90
Economic resources
Does not own shoes	36 (75.00)	70 (68.63)	.22 (.40)	1.25	[.56, 2.75]	.59
Does not own two sets clothes	7 (14.58)	20 (19.61)	−.45 (.49)	0.64	[.24, 1.68]	.36
Has lacked school fees	29 (60.42)	63 (61.76)	.09 (.36)	1.10	[.54, 2.23]	.80

Note. OR = odds ratio; CI = confidence interval.

Hypothesis 2

Orphan status did not moderate relationships of any of the psychosocial or economic variables with history of sexual activity. Orphan status did, however, moderate relationships between several psychosocial and economic variables, with sex-related self-efficacy. Table V shows these significant moderator effects for self-efficacy. Effect sizes for all moderation effects were small.

Table V.

Significant Moderator Effects of Orphan Status on Relationships of Mental Health, Social Support, and Economic Resources Variables with Sex-Related Self-Efficacy

Regression Step	Predictor	R	Adjusted R2	ΔR2	p
Step 1	Emotional problems	.074	−.004	.006	.62
Step 2	OS	.085	−.005	.002	.45
Step 3	Emotional problems × OS	.149	.007	.015	.03*
Step 1	PTSD: avoidance	.116	.004	.013	.23
Step 2	OS	.132	.005	.004	.26
Step 3	PTSD: avoidance × OS	.176	.016	.014	.04*
Step 1	P–Y communication barriers	.242	.050	.058	.00**
Step 2	OS	.244	.048	.001	.54
Step 3	P-Y communication barriers × OS	.310	.082	.036	.00**
Step 1	HH income	.035	−.008	.001	.94
Step 2	OS	.079	−.006	.005	.21
Step 3	HH income × OS	.171	.014	.023	.01*

Notes. All models controlled for youth age and gender. PTSD = posttraumatic stress disorder; P–Y = parent–youth. HH = household; OS = orphan status.

*p < .05, **p < .01.

Orphan status moderated the relationship between emotional problems and sex-related self-efficacy. For single and double orphans, but not nonorphans, emotional symptoms had an inverse, though nonsignificant, relationship with self-efficacy. For avoidance symptoms of PTSD, among nonorphans, but not single or double orphans, there was an unexpected positive association of avoidance symptoms with self-efficacy. Orphan status also moderated the relationship between barriers to parent–youth communication and self-efficacy such that barriers to communication had a stronger inverse relationship with self-efficacy among orphans than nonorphans; relationships were nonsignificant for all groups. Household income had a positive relationship with self-efficacy for orphans, but not nonorphans, though these relationships were nonsignificant across groups.

Orphan status also moderated the relationship between monthly income and beliefs related to risk behavior (R = .29, adjusted R²= .07, ΔR²= .029, p < .002). For nonorphans, but not for single or double orphans, lower income was related to beliefs indicating acceptance of risky behavior. This did not support the hypothesis that lower income would have a stronger effect on orphans.

Discussion

This study builds on the small body of literature on orphans’ HIV risk, social support, economic resources, and mental health. In contrast to previous studies, in this sample, orphans and nonorphans did not differ on rates of sexual activity, beliefs about sex, or sex-related self-efficacy. It is possible that in this community with such high HIV prevalence, risk among all youth may be elevated such that losing a parent does not significantly increase risk. Further, in this community, transactional sex associated with fishing communities is common, which may result in social norms and expectations that are more accepting of sexual risk behavior. These factors may be more influential as risk factors than losing a parent. Additionally, it is possible that although orphans were no more likely to exhibit increased risk in this age range, they could be at higher risk as adults.

Although orphans did not exhibit more indicators of HIV risk, they did report more mental health problems, lower social support, and less access to economic resources. Consistent with previous studies (Cluver & Gardner, 2007), orphans reported more symptoms of emotional problems, including depression and some symptoms of traumatic stress. They reported lower levels of support from caregivers and others. Consistent with Bicego et al. (2003), we found no income differences in households with orphans, though orphans were more likely to lack basic material needs than nonorphans. Taken together, this suggests that within households, fewer resources may be allocated for orphans (Juma et al., 2007, Nyambedha et al., 2003). In this study, double orphans showed more severe problems than single orphans in some domains. Only double orphans exhibited more exposure to trauma, more traumatic stress, and less social support specifically from caregivers.

Despite the fact that HIV-related behaviors and beliefs did not differ by orphan status, moderation analyses indicate that some psychosocial factors may be more influential on orphans’ level of self-efficacy for engaging in preventative behaviors such as refusing unwanted sex and using condoms. Moderator analyses supported the hypothesis that among orphans, emotional problems and poorer communication with caregivers about sex were more strongly associated with lower self-efficacy than for nonorphans. Effects were small, but warrant further investigation, as targeting psychosocial well-being of orphans could be important for HIV prevention. In such interventions, targeting the quality and content of caregiver communication about sex is likely important; higher levels of social support were associated with increased HIV risk in this sample in a previous study (Puffer et al., 2011), suggesting that in this context, caregivers and other adults may view early sexual activity as normative (especially for males).

Among orphans, results suggest that risk for mental health problems and lower psychosocial support may increase over time. Those whose parents have been deceased longer exhibited more conduct problems, symptoms of traumatic stress, and barriers in communication with caregivers about sex and HIV. These results did not support our hypothesis that recent parental death would result in short-term grief and adjustment difficulties. Interestingly, conduct problems and caregiver communication difficulties were not related to orphan status in other analyses. Rather, within orphans, these difficulties may be more related to the cumulative time effect of being orphaned. Though this study did not examine the mechanisms underlying this, a potential explanation is that there could be a sensitive period earlier in development during which losing a parent leads to increased difficulties during adolescence. For instance, children who lose parents during early- or middle-childhood may have difficulty forming secure attachments or coping with stress; these difficulties could interfere with an adolescent’s ability to communicate openly with caregivers or to control their behavior. Another potential explanation is that caregivers and other adults provide more support and comfort during the period immediately following parental death but this decreases over time.

Limitations and Future Directions

This study included a relatively small sample and used a cross-sectional design, limiting generalizability. The correlational design precludes determinations of causality, and the small number of double orphans hindered our ability to detect small differences among types of orphans. Additionally, as many younger adolescents did not report sexual activity, power was too low to examine risk behaviors such as condom use and number of partners. Also, although the use of random selection is a strength, our sampling frame included only adolescents enrolled in school. This is tempered by data documenting that 97% of all children in Nyanza attend school, and the fact that the large percentage of orphans in our school-based sample (45%) exceeds the estimated overall percentage of orphans in the province (35%; Juma et al., 2007). Finally, some measures in this study that were adapted from scales developed using Western populations showed lower reliabilities, potentially limiting their meaningfulness in this context. Future validation studies are necessary to improve existing assessment tools.

Future studies should examine orphan well-being in larger samples, and longitudinal prospective studies are needed to examine causal relationships between mental health, social, and economic consequences of parental death and risk behavior. Particularly for understanding how time since parental death affects orphans’ well-being, longitudinal studies are necessary to capture the emotional and environmental changes that occur for children across different developmental stages.

Conclusions

A main goal of this study was to contribute to literature informing how governmental and nongovernmental organizations should focus their efforts in developing and disseminating HIV prevention interventions. Results are mixed. Finding no differences in HIV risk indicators among orphans lends partial empirical support to the argument made by some organizations that all children, not only orphans, in low resource settings should be targeted for HIV prevention. On the other hand, results suggest that orphans are more likely than nonorphans to need mental health and social support interventions. Further, the results of moderation analyses suggest that psychosocial problems might put orphans at higher risk for HIV than their nonorphan peers. Thus, interventions for orphans designed to prevent and treat internalizing symptoms could not only be useful to improve psychological well-being of orphans, but also could ultimately have a protective effect on HIV risk as well. Family or community-level interventions could be useful to (a) increase the capacity of caregivers to provide adequate social support and (b) assist families in managing finances and increase awareness about the importance of more equitable distribution of resources.

Results are consistent with previous studies showing more mental health and economic difficulties among orphans, and this study is among the first to document that orphans experience lower social support. Results contribute to the literature on HIV risk among orphans by demonstrating that there may not be a direct link between orphan status and risk behavior across all contexts and age ranges; in communities with high rates of HIV and norms more accepting of risky sexual behavior, risk for all youth may be elevated such that being an orphan is a less distinct risk factor.

Further research is needed to understand the complex ways in which developmental and health outcomes of orphans are determined. It is likely that orphans’ physical, mental, and interpersonal well-being is influenced by interactions between the timing and circumstances of parental death, characteristics of their relationship(s) with caregivers, and their community context. Results of this and future studies on orphan adjustment in SSA are important for informing interventions in communities coping with increasing numbers of orphans and persistently high rates of HIV.

Funding

This project was funded in part by the Duke Global Health Institute, Johnson and Johnson Corporation, and a 2009 developmental grant from the Duke University Center for AIDS Research (CFAR) funded by the NIH (P30 AI64518). This work also was supported by the National Institutes of Health Office of the Director, Fogarty International Center, Office of AIDS Research, National Cancer Center, National Eye Institute, National Heart, Blood, and Lung Institute, National Institute of Dental & Craniofacial Research, National Institute On Drug Abuse, National Institute of Mental Health, National Institute of Allergy and Infectious Diseases Health, and NIH Office of Women's Health and Research through the International Clinical Research Scholars and Fellows Program at Vanderbilt University (R24 TW007988) and the American Relief and Recovery Act.

Conflicts of interest: None declared.