Gender Differences in Social Support Vary by Environment Among Older People With HIV in South Africa and Uganda

Abstract

Background and Objectives

Sub-Saharan Africa is home to 3.7 million older adults living with HIV, who experience high rates of comorbid conditions. Formal services other than HIV clinical care are largely unavailable. Overall, women are the mainstay of informal social support networks, and older women with HIV may face burdens due to family caregiving expectations. Thus, it is important to understand the extent of informal support provided to older adults living with HIV, and how this is affected by gender.

Research Design and Methods

We examined differences in social networks, needs, social support and caregiving, and perceptions of support adequacy among women and men aged 50 and older living with HIV in Uganda (n = 101) and South Africa (n = 108), mostly rural and suburban populations, respectively. We used multiple regression to determine whether there was an association between gender and the amount of social support received and whether that varied by research site.

Results

Men were more likely than women to receive support from a partner. Women were more likely to live with offspring, both providing and receiving care. In South Africa but not Uganda, women received more help from family than men did. There was no gender difference in getting help from friends, but it was more common in Uganda. Living alone was strongly associated with less family help and more help from friends.

Discussion and Implications

Older women with HIV in sub-Saharan Africa tend to be more heavily involved in social support exchanges—both providing and receiving care—than their male peers, but place matters. Interdependence is high in rural Uganda, where formal services are scarce and needs exceed resources. Given the projected growth in this population, stronger formal supports are needed for communities and older people with HIV, especially those who live alone.

Translational Significance: Sub-Saharan Africa is home to 3.7 million older adults with HIV, with high rates of comorbid conditions. In the absence of formal services other than HIV care, essential informal support networks often rely on women, but women need support too. This study found that in South Africa, older women with HIV rely on family more than men do. In Uganda, interdependence is high for both men and women, but unmet needs are also high. Older people with HIV, especially women, continue to provide care. Policies that help older adults with HIV are likely to help their communities, and vice versa.

Background and Objectives

With effective antiretroviral therapy now widely available in sub-Saharan Africa, the number of people aged 50 and older living with HIV in the region is nearing 4 million (Autenrieth et al., 2018). This is a success story, but aging brings increasing rates of comorbid conditions in this population, such as hypertension and stroke (Matlho et al., 2022; Negin et al., 2012). Many areas of sub-Saharan Africa lack a strong infrastructure for health and social services (Small et al., 2019). Traditionally, families care for their older relatives, but urban migration and HIV have affected these social roles (Golaz et al., 2017; Schatz & Seeley, 2015; Schatz et al., 2015).

Social support is important for the well-being of older adults with and without HIV, and it may also bolster treatment adherence and retention in HIV care (Knight & Schatz, 2022). Older adults need varying degrees of financial, physical, or instrumental, and emotional support. They also provide a substantial amount of support to others, particularly by caring for grandchildren in the absence of parents who have died from AIDS or other causes, or who have migrated for work (Golaz et al., 2017; Schatz, 2007; Schatz & Seeley, 2015). This care work most often falls to women, who are frequently without spousal support in old age (Harling et al., 2020; Mugisha et al., 2013).

This article aims to better understand the role of gender in social networks and social support for older adults with HIV in sub-Saharan Africa, and whether that role differs between urban/suburban South Africa and rural Uganda. Sub-Saharan Africa is not homogenous, and people and cultures are continually adapting to economic and environmental conditions (Hall & Posel, 2019).

Differences in Support in South Africa and Uganda

South Africa has a more urban population, a higher average income, and more formal employment than many other countries in the region, including Uganda. Its workforce is largely in the service sector, while Uganda’s is primarily agricultural (United Nations, 2017). South Africa also has a widespread social grant/pension system (Schatz et al., 2015). Uganda has introduced a modest Senior Citizens’ Grant that has been found to help reduce poverty, but many are left out; as of 2021, the eligibility age was 80 (Emuk, 2022; Gelders & Athias, 2019). South Africa has a stronger infrastructure to support older adults, although health systems may be poorly organized even in urban areas (Nxumalo et al., 2016; Small et al., 2019). In the absence of formal support, family support is especially important for older adults.

Older adults in rural Uganda often report fairly large social networks; in villages, neighbors are often relatives (Golaz et al., 2017). Yet these networks may be short on economic resources. In South Africa, labor migration under apartheid laws left a legacy of familial separation (Hall & Posel, 2019).

In an earlier analysis of the data set used in the present article, we examined the social networks of adults aged 50 and older living with HIV in Uganda and South Africa and found significant differences (Brennan-Ing et al., 2022). Compared with the older adults in South Africa, those in Uganda received less instrumental help but more emotional support from their families. They also had greater needs: substantially more physical health conditions and more depressive symptoms. Study participants in Uganda had larger families and were more likely to be living with and caring for their children and/or grandchildren than their peers in South Africa. Overall, participants in Uganda described more help from friends and mutual assistance in the community, likely due in part to a high level of poverty and a lack of formal services (Brennan-Ing et al., 2022; Rishworth & Elliott, 2022).

Gender, Social Support, and Social Exchange

The analysis of social support reported in Brennan-Ing and colleagues (2022) raised additional questions, notably how support differed by gender in the two samples. Given that men in sub-Saharan Africa often marry younger women, and women’s life expectancy is longer than men’s, older men with HIV are much more likely than older women with HIV to be married and receive care from a spouse or partner (Harling et al., 2020; Mugisha et al., 2013). In patriarchal societies, a woman may be cutoff from her husband’s family and their children after his death or a divorce (Golaz et al., 2017). If women who are widowed or divorced receive less family support than men do, they may rely more on friends or neighbors, though there is typically a strong preference for support from family (Knight & Schatz, 2022; MacPhail et al., 2022). In a qualitative study of older adults with HIV in South Africa, both men and women said they received help from adult children, but men also relied heavily on partners, whereas women got additional help from other family members, such as siblings and mothers (MacPhail et al., 2022). Labor migration also has a gender dimension; in South Africa, men were more likely to migrate to urban areas, living apart from their families, whereas women’s labor migration was closer to home, enabling them to uphold caregiving responsibilities (Camlin et al., 2014).

Schatz and Seeley (2015) have suggested a conceptual framework of gendered identities and social exchange to examine care provision in sub-Saharan Africa. Over the course of their lives, older adults have developed a sense of who they are and should be at a given age, a sense that is shaped by kinship structures, family relationships, and social and economic conditions. Often, according to traditional gender roles, men provide material support, whereas women provide personal care to partners, parents, children, and grandchildren in the home (Mugisha et al., 2013; Schatz, 2007; Schatz & Seeley, 2015). MacPhail and colleagues (2022) found, for example, that older women living with HIV in South Africa reported more responsibility for adult children and grandchildren than their male counterparts did. Men viewed their family role as providing financial support, and they accepted that their capacity to do so had diminished with age.

Particularly in societies that value interdependence, however, both older women and men may need to feel and show they are still useful (Freeman, 2016; Mugisha et al., 2013). In South Africa, receipt of the old age pension can make an older adult a desired household member, “productive” rather than “dependent” (Madhavan et al., 2017; Schatz et al., 2015). Being able to provide one kind of service, such as childcare or financial support, may increase one’s likelihood of receiving a different kind, such as help carrying water or firewood, or receiving care later if their health declines (Kasedde et al., 2014; Reynolds et al., 2022). Indeed, older adults sometimes request grandchildren to live with them and help them in return for care (Reynolds et al., 2022; Schatz & Seeley, 2015; Small et al., 2019). Parents may also emphasize the lifelong care they have given their children to reinforce their right to receive care (Freeman, 2016; Schatz & Seeley, 2015). Nevertheless, caregiving, especially caring for more than one person or when resources are limited, can be a substantial burden for older adults (MacPhail et al., 2022; Mugisha et al., 2013; Schatz, 2007).

This article is an exploratory analysis following up on findings reported in Brennan-Ing et al. (2022), asking: (1) Do men and women in Uganda and South Africa have different family relationships and social networks? (2) Does gender affect the levels or kinds of support received from family members and friends in these populations? (3) Do older adults who provide care to grandchildren or adults receive more support from family members?

Research Design and Methods

Study Sample

This study used cross-sectional data from the Research on Older Adults with HIV (ROAH) Africa project (2013–2015), with 209 participants in Uganda and South Africa (Brennan-Ing et al., 2022; Negin et al., 2016). Individuals aged 50 or older, with an HIV-positive serostatus, were eligible.

Researchers in Uganda recruited participants (n = 101) from individuals enrolled in the World Health Organization’s Health and Wellbeing of Older People Study. Participants were living in southwestern Uganda, in the rural Kalungu district, or around Entebbe in the Wakiso district (Nyirenda et al., 2013). Most were of Baganda ethnicity. Participants in South Africa (n = 108) were recruited from an antiretroviral therapy clinic in central Johannesburg. The most common ethnicities were Tswana (44%) and Zulu (22%). Although the Baganda, Tswana, and Zulu employ different complex kinship structures, all three are patriarchal and clan-based (see Brennan-Ing et al., 2022, for additional description of these social systems). Despite increasing forces putting distances between family members, clan ties and traditions persist (Seeley, 2015).

Interviewers administered a survey adapted from the U.S. ROAH study (Karpiak & Brennan, 2009) for sub-Saharan African populations. It was translated from English at the research sites into the local languages of the participants (in Uganda, Luganda; in South Africa, isiZulu and isiSotho). The survey covered demographics, mental health, physical health, social networks, HIV treatment experience, health-related quality of life, and sexual behavior. ROAH Uganda was approved by the Science and Ethics Committee of the Uganda Virus Research Institute and the Uganda National Council for Science and Technology. ROAH South Africa was approved by the institutional review board of the University of New England (Australia) and the Human Subjects Ethics Committee of the University of the Witwatersrand (South Africa). Details on the sample recruitment and measures have been published elsewhere (Brennan-Ing et al., 2022; Negin et al., 2016).

Measures

Social network

Participants were asked whether they had any children, grandchildren or great-grandchildren, siblings, other relatives they were in contact with, close friends, and neighbors known well. If they did, they were asked how many of each, and how often they saw at least one of these people face-to-face or spoke with them on the telephone, using a 5-point ordinal scale ranging from “daily” to “once per year or less often.” Having at least monthly in-person contact or at least weekly phone contact with at least one person constituted having a functional support element (e.g., functional child); a person could have a maximum of five, as neighbors were not included (Cantor & Brennan, 2000). Participants were asked how close they felt to their children and other family members on a 4-point scale ranging from “very close” to “not close at all.”

Social exchange

We coded participants as caregivers if they said they were the ones mainly responsible for caring for a grandchild, a great-grandchild, or another child, or if they were currently caring for an adult relative or friend. Employment responses included working full-time, working part-time, unemployed, retired, self-employed, on a pension/social grant, disabled, or other (write-in). We considered those working full-time or part-time, or self-employed as employed.

Needs and support

Participants were asked, “Have you ever needed physical (financial, emotional) help because of HIV/AIDS?” and if so, who provided that help. Support availability and adequacy were measured by questions about whether the participant had someone to count on for help with tasks of daily living (instrumental help) and emotional support and how much extra help was needed.

Assistance from family and friends

Participants were asked whether they received instrumental and emotional support in eight categories from family and from friends or neighbors, and how often in the past month, on a 6-point scale ranging from “not at all or occasionally” to “every day.” The responses have been collapsed into four categories, “daily,” “several times per week,” “weekly/monthly,” and “not at all/occasionally.” The number of ways family or friends helped is the sum of types of assistance provided at least once per month.

Negative interactions

Participants were asked how often family members or friends/neighbors were reluctant to talk, upset them or hurt their feelings, or refused to help when asked in the past month, on a 6-point scale ranging from “not at all or occasionally” to “every day.”

Covariates

Demographic characteristics included age in years, gender (female/male), marital status, household composition, and geographic area of residence: city, suburb, town, trading center, or village. All health variables were self-reported, including receipt of an AIDS diagnosis. Self-rated health was on a 5-point scale. Comorbid conditions in addition to HIV represent responses to questions about any of 26 medical conditions, including “other condition.” Depressive symptoms were measured with the 10-item version of the Center for Epidemiologic Studies Depression scale (CES-D 10; Andersen et al., 1994; Radloff, 1977). Higher scores indicate a higher sum of depressive symptoms (range = 0 to 30). Internal consistency for the CES-D 10 was acceptable; Cronbach’s alpha = .72 in the combined sample, .74 in South Africa, and .70 in Uganda. Correlations between CES-D score and number of comorbidities (.571, p < .001) and loneliness (.267, p < .001) indicate convergent validity for the measures in the study population.

Design and Statistical Analysis

This study used a cross-sectional survey design. Differences in data between genders at each of the research sites were evaluated using chi-square tests for categorical variables and one-way ANOVA for continuous variables. We used p < .05 as our criterion for significant differences. We conducted ordinary least squares multiple regression analyses to determine whether there was an association between gender and the number of types of social support received and whether that varied by research site. Listwise deletion of missing data was used for analysis. We used SPSS v.27 for all analyses.

Results

Sample Characteristics

As noted in the introduction, the two samples, from South Africa and Uganda, differ demographically in a number of ways. (Some comparisons not involving gender were first reported in Brennan-Ing et al., 2022.) In Uganda, 90% of the sample reported living in a village, and 93% reported working, usually informally or through self-employment. In South Africa, 84% reported suburban or urban residence and half the sample worked full-time or part-time, whereas one-third were retired or receiving a pension. The South Africa sample was younger (mean age 58, compared with 61 in Uganda) and 72% were female, compared with 58% of the Uganda sample.

Demographic differences between genders were similar in the two samples but more pronounced, and statistically significant, in Uganda (Table 1). The men in the sample were older, on average, than the women, a significant difference of 4.1 years in Uganda. Men were much more likely to be partnered and less likely to be widowed, especially in Uganda, where 60% of men were married or cohabitating and 29% widowed, compared with 14% partnered and 58% widowed among the women. Yet one in five men in the Uganda sample lived alone, compared with just one out of 59 women. Women in both locations were more likely to live with grandchildren, with or without an adult child in the household. (All references from study data to grandchildren include great-grandchildren.) There were no significant gender differences in self-rated health, number of comorbid conditions, or depressive symptoms at either site. In the Uganda sample, however, women were significantly more likely than men to report an AIDS diagnosis (68%, compared with 46% of men).

Table 1.

Household Characteristics and Health by Gender in South Africa and Uganda Samples

Characteristic	Total		South Africa		Uganda
	Women (n = 137)	Men (n = 72)	Women (n = 78)	Men (n = 30)	Women (n = 59)	Men (n = 42)
Age, mean (SD)	58.5 (6.34)	61.69 (7.99)**	57.87 (5.95)	59.27 (6.42)	59.32 (6.79)	63.43 (8.59)**
Marital status, %						***
Married	14.7	43.1	19.5	43.3	8.5	42.9
Widowed	41.2	23.6	28.6	16.7	57.6	28.6
Divorced/separated	22.8	15.3	18.2	20.0	28.8	11.9
Cohabitating	6.6	12.5	7.8	6.7	5.1	16.7
Cowife	0.7	0	1.3	0	0	0
Single/never married	14.0	5.6	24.7	13.3	0	0
Living situation, %
Alone	13.9	23.9	23.1	30.0	1.7	19.5**
With adult child	46.3	41.7	38.5	16.7*	56.9	59.5
With grandchild	52.9	27.8***	37.2	13.3*	74.1	38.1***
With both child and grandchild	30.9	12.5**	23.1	3.3	41.4	19.0**
Employment status, %
Employed	70.6	70.4	52.6	43.3	94.8	90.2
Retired/on pension	16.9	18.3	28.2	36.7	1.7	4.9
Self-rated health, %
Excellent	6.7	8.3	1.3	0	13.6	14.3
Good	34.1	45.8	27.6	40.0	42.4	50.0
Fair	47.4	33.3	60.5	53.3	30.5	19.0
Poor/very poor	11.9	12.5	10.5	6.7	13.6	16.7
AIDS diagnosis, %	38.5	30.4	15.8	10.0	67.8	46.2*
Comorbid conditions, mean (SD)	3.07 (2.87)	3.47 (3.31)	1.27 (1.04)	1.10 (1.09)	5.46 (2.78)	5.17 (3.33)
CES-D depressive symptoms, mean (SD)	10.12 (5.93)	9.22 (6.58)	7.94 (3.93)	6.43 (4.80)	13.00 (6.86)	11.21 (7.00)

Notes: N = 209. SD = standard deviation. Due to missing values, the number of participants varies by characteristic from 204 to 209. Comparison of gender within each site uses one-way ANOVA for continuous variables and the Pearson chi-square test for categorical variables.

^* p < .05.

^** p < .01

^*** p < .001.

Gender Differences in Social Networks

Women had stronger links to children and grandchildren (Table 2), especially in South Africa, where men were less likely to report having a child or grandchild they talked to often (60% and 53%, respectively, vs 92% and 76% of women). Women were more likely to be caregivers of children or adults, significantly so in South Africa (41% of women vs 10% of men). In Uganda, high percentages of both men and women reported feeling close to their children and grandchildren, and to a lesser degree other kin, but there was a striking gender difference in South Africa. Although nearly all the women reported feeling “very close” to their children and grandchildren (99% and 97%, respectively), just 54% of men with children and 68% of men with grandchildren said they felt very close to them. Three-quarters of the women in South Africa reported feeling very close to their siblings, compared with about half of the men.

Table 2.

Social Networks and Caregiving by Gender in South Africa and Uganda Samples

Social network characteristic	N	South Africa		Uganda
		Women (n = 78)	Men (n = 30)	Women (n = 59)	Men (n = 42)
Social network components, % (n)
Spouse/partner	209	28.2 (22)	50.0 (15)*	13.6 (8)	59.5 (25)***
Functional child	209	92.3 (72)	60.0 (18)***	83.1 (49)	81.0 (34)
Functional grandchild or great-grandchild	209	75.6 (59)	53.3 (16)*	79.7 (47)	71.4 (30)
Functional sibling	208	52.6 (41)	53.3 (16)	47.5 (28)	43.9 (18)
Functional other relative	208	51.3 (40)	43.3 (13)	44.1 (26)	41.0 (16)
Functional friend	208	39.7 (31)	46.7 (14)	67.8 (40)	78.0 (32)
Neighbor known well	199	70.7 (53)	58.6 (17)	98.2 (54)	92.5 (37)
Neighbors help each other “all of the time”	195	19.4 (13)	25.0 (7)	72.9 (43)	75.6 (31)
Caregiver of child or adult	204	40.8 (31)	10.0 (3)**	67.8 (40)	53.8 (21)
In general, feel very close to,a % (n)
Children	194	98.6 (70)	53.8 (14)***	84.2 (48)	82.5 (33)
Grandchildren	180	96.9 (63)	68.0 (17)**	75.4 (43)	81.8 (27)
Siblings	184	74.6 (53)	48.3 (14)***	61.5 (32)	65.6 (21)
Other relatives	132	74.4 (32)	66.7 (10)	63.6 (28)	73.3 (22)
Size of network components, mean (SD)	209
Living children		2.69 (1.72)	3.23 (2.11)	6.12 (3.68)	8.32 (7.43)
Living grandchildrenb		3.14 (3.01)	3.10 (2.93)	9.80 (6.59)	9.05 (7.26)
Living siblings		3.97 (2.09)	3.80 (2.14)	3.61 (4.09)	3.90 (4.69)
Other relatives		1.06 (1.29)	1.20 (1.71)	3.44 (4.44)	4.07 (5.13)
Friends		0.90 (2.39)	1.50 (2.49)	1.78 (1.78)	2.51 (2.46)
Neighbors known well		2.17 (2.45)	1.27 (1.62)	2.83 (1.71)	2.88 (1.73)
Functional support elements		2.47 (1.11)	2.47 (1.11)	2.53 (1.07)	2.85 (1.31)
Social network		14.22 (6.21)	14.60 (5.31)	27.71 (10.83)	31.32 (14.07)

Notes: Total N = 209. SD = standard deviation. Comparisons of gender within each site use one-way ANOVA for continuous variables and the Pearson chi-square test for categorical variables.

^aAmong participants who have these living family members.

^bIncludes grandchildren and great-grandchildren.

^* p < .05.

^** p < .01

^*** p < .001.

Gender Differences in Needs and Support

As Table 1 indicated, participants in Uganda faced greater health challenges than those in the South Africa sample, and a greater share of those in Uganda reported needing help (Table 3). There were no significant gender differences in the need for physical or emotional help because of HIV. Men in South Africa were more likely than women to say they needed financial help currently or in the past; 65% of women said they had never needed financial help because of HIV, compared with 40% of men. In Uganda, the opposite held true, with just 21% of women saying they had never needed financial help, compared with 45% of men.

Table 3.

Needs and Support by Gender in South Africa and Uganda Samples

Response	N	South Africa		Uganda
		Women	Men	Women	Men
		(n = 78)	(n = 30)	(n = 59)	(n = 42)
		% (n)	% (n)	% (n)	% (n)
Need physical help	208
Yes, currently		6.4 (5)	13.3 (4)	39.7 (23)	40.5 (17)
Not now, needed in the past		47.4 (37)	30.0 (9)	22.4 (13)	7.1 (3)
No, never needed		46.2 (36)	56.7 (17)	37.9 (22)	52.4 (22)
Need financial help	207		*		*
Yes, currently		9.0 (7)	20.0 (6)	54.4 (31)	47.6 (20)
Not now, needed in the past		25.6 (20)	40.0 (12)	24.6 (14)	7.1 (3)
No, never needed		65.4 (51)	40.0 (12)	21.1 (12)	45.2 (19)
Need emotional help	209
Yes, currently		10.3 (8)	3.3 (1)	20.3 (12)	33.3 (14)
Not now, needed in the past		55.1 (43)	43.3 (13)	39.0 (23)	23.8 (10)
No, never needed		34.6 (27)	53.3 (16)	40.7 (24)	42.9 (18)
Who provided needed helpa
Spouse/partner	181	5.2 (3)	18.2 (4)	5.1 (3)	33.3 (14)***
Adult child	181	34.5 (20)	13.6 (3)	44.1 (26)	19.0 (8)*
Sibling	180	22.8 (13)	22.7 (5)	5.1 (3)	7.1 (3)
Other family member	181	46.6 (27)	22.7 (5)	11.9 (7)	9.5 (4)
Friend/neighbor	180	10.5 (6)	13.6 (3)	10.2 (6)	16.7 (7)
Instrumental help availabilityb	206
All or most of the time		38.7 (29)	36.7 (11)	40.7 (24)	38.1 (16)
Not at all		24.0 (18)	26.7 (8)	33.9 (20)	33.3 (14)
Instrumental help adequacyc	192
Got all needed		61.3 (46)	80.0 (24)	26.4 (14)	41.2 (14)
Need a lot more		4.0 (3)	3.3 (1)	34.0 (18)	35.3 (12)
Emotional support availabilityb	206
All or most of the time		11.8 (9)	3.3 (1)	71.2 (42)	53.7 (22)
Not at all		30.3 (23)	46.7 (14)	11.9 (7)	9.8 (4)
Emotional support adequacyc	201				**
Got all needed		59.2 (45)	86.7 (26)	36.8 (21)	57.9 (22)
Need a lot more		6.6 (5)	0	19.3 (11)	34.2 (13)

Notes: Comparisons of gender within each site use the Pearson chi-square test of significance.

^aResponses from those who said they needed or had needed help. Categories are not exclusive.

^bResponses not shown: some of the time; only occasionally.

^cResponses not shown: need some more; need a little more.

^* p < .05.

^** p < .01

^*** p < .001.

When they needed help, men were more likely than women to receive it from a spouse (33% vs 5%, respectively, in Uganda), and women were more likely than men to receive it from their children (44% vs 19%, respectively, in Uganda); these differences were significant in Uganda and in the combined sample. Gender differences were not significant in instrumental help availability and adequacy, or in emotional support availability. On the question of emotional support adequacy, men were more likely than women to say they received all they needed (87% of men in South Africa and 58% of men in Uganda, vs 59% of women in South Africa and 37% of women in Uganda), but at the same time, men in Uganda were more likely than women to say they needed a lot more (34% and 19%, respectively).

Figure 1 shows the frequency with which women and men received help from family members in various ways. Shopping and housework or cooking were the most frequent forms of help. For example, in South Africa, 33% of women and 44% of men had help with housework or cooking every day; in Uganda, 41% of women and 39% of men received housework or cooking help every day. In South Africa, women received more family support than men in several areas: family members helped participants at least monthly with managing money or paying bills (48% of women and 18% of men), giving advice on big decisions (70% of women and 39% of men), and talking when the participant was feeling low (65% of women and 29% of men) or about personal matters (55% of women and 26% of men). There were no significant differences in the amount of help women and men in Uganda received.

Figure 1.

Frequency of help from family members. South Africa: women, n = 78, men, n = 30; Uganda: women, n = 59, men, n = 42. Due to missing values, the number of participants varies by characteristic, from 204 to 205. Comparisons of gender within each site use the Pearson chi-square test of significance. *p < .05 **p < .01.

As Figure 2 indicates, participants in Uganda received help, especially instrumental help, from friends and neighbors more often than participants in South Africa. About half the men and women in both locations received emotional support from friends at least monthly. Notably, 52% of men in South Africa sought advice from friends and 48% talked with friends, suggesting a greater reliance on friends than family for emotional support. In Uganda, men were significantly more likely than women to report friends helping with correspondence and relied on friends when feeling low more often than women did.

Figure 2.

Frequency of help from friends or neighbors. South Africa: women, n = 78, men, n = 30; Uganda: women, n = 59, men, n = 42. Due to missing values, the number of participants varies by characteristic, from 185 to 199. Comparisons of gender within each site use the Pearson chi-square test of significance. *p < .05 **p < .01.

In Uganda, negative interactions with family members and friends were uncommon for men and women (Supplementary Table S.1). In South Africa, about half the women said family members upset them at least once a month (49%, vs 29% of men; p = .066), but 39% of men said family members were reluctant to talk at least once a month, compared with 29% of women (p = .315).

Regression Analysis of Ways Family and Friends Helped

To determine which factors besides gender are associated with the kinds of help that family and friends provided to older adults with HIV in South Africa and Uganda, we fit a regression model with the measures of health, social connectedness, and need that showed at least a moderate bivariate correlation with the number of ways family members helped an individual. Pension receipt was not a significant factor. We performed the final analysis with the following factors in Model 1: number of comorbid conditions, CES-D score, employment status, caregiver status, living alone or not, number of functional support elements, and number of ways friends help. In Model 2, we added gender (female) and research site (Uganda). To test whether association with gender was moderated by research site, we added an interaction term for gender and site in Model 3. For ease of interpretation, we report unstandardized coefficients in the text; the tables include standardized coefficients for comparison.

Table 4 presents the results of the regression analysis on number of ways family members helped. In Model 1, the number of comorbid conditions was negatively associated with family help (B = –0.175, p = .017); depressive symptoms had only a slight association. Employment was associated with nearly one fewer way of receiving family help (B = –0.886, p = .029), and living alone had a substantial negative association (B = –2.825, p < .001). Caregiving was not a significant factor. Receiving help from friends had a positive association with family help (B = 0.226, p = .002). After adding gender and research site, in Model 2, there was no longer a significant association with comorbid conditions, depressive symptoms, or employment (all more common in Uganda), but the coefficients for living alone and receiving help from friends remained significant. Gender had a positive but nonsignificant association with family help, and being in the Uganda sample had a negative but nonsignificant association.

Table 4.

Factors Associated With the Number of Ways Family Members Helped

Variable	Model 1			Model 2			Model 3
	B	SE	β	B	SE	β	B	SE	β
Constant	3.827***	0.546		3.326***	0.602		2.771***	0.629
Comorbid conditions	–0.175*	0.073	–0.194	–0.077	0.085	–0.085	–0.081	0.084	–0.090
CES-D	0.068*	0.034	0.151	0.056	0.034	0.125	0.059	0.033	0.132
Employed	–0.886*	0.402	–0.147	–0.659	0.441	–0.109	–0.649	0.434	–0.107
Caregiver	0.641	0.373	0.116	0.547	0.379	0.099	0.473	0.374	0.086
Lives alone	–2.825***	0.483	–0.394	–2.899***	0.498	–0.404	–3.034***	0.493	–0.423
Functional support elements	0.283	0.159	0.117	0.299	0.159	0.124	0.279	0.157	0.116
Ways friends help	0.226**	0.072	0.201	0.273***	0.074	0.243	0.259**	0.073	0.231
Female				0.627	0.373	0.107	1.559**	0.508	0.267
Research site: Uganda				–0.851	0.531	–0.154	0.401	0.704	0.073
Female × Uganda							–1.845**	0.695	–0.304
R 2	0.307			0.330			0.354
Adjusted R2	0.282			0.298			0.320

Notes: Listwise N = 201.

CES-D = Center for Epidemiologic Studies Depression scale; SE = standard error.

^* p < .05

^** p < .01

^*** p < .001.

Adding the interaction of gender and research site, in Model 3, greatly clarified the results. Living alone and receiving help from friends remained significant. Living alone was associated with receiving family help in three fewer ways, on average, holding other factors constant (B = –3.034, p < .001). For each kind of help received from friends, there was a 0.26 increase in kinds of help from family (B = 0.259, p = .001). The association between female gender and family help rose sharply (B = 1.559, p = .002); the association between site and family help changed directions and lost significance; and the interaction of female gender and being in the Uganda sample was negatively associated with family help (B = –1.845, p = .009). These results indicate that there is a strong link between gender and the ways family members help, but only in South Africa, as the coefficients cancel out for women in Uganda. After controlling for other factors, women in South Africa, received help from family in about 1.6 more ways than men did, on average, but women and men in rural Uganda received about the same amount of family help. This is consistent with the descriptive data in Figure 1.

The regression analysis for the number of ways friends and neighbors help used identical factors, except that “number of ways family helps” replaced “number of ways friends help” (Table 5). In this analysis, in Model 1, employment was positively associated with help from friends (B = 0.876, p = .027), as were living alone (B = 1.199, p = .018), number of functional support elements (B = 0.310, p = .046), and number of ways family members help (B = 0.215, p = .002). When gender and site were added in Model 2, living alone and receiving family help remained significant, but working for pay lost significance and functional support elements dropped to marginal significance (B = 0.295, p = .051). Gender had no relationship with help from friends, but being in the Uganda sample was associated with receiving two more kinds of help from friends (B = 2.033, p < .001). Including the interaction term for gender and research site (Model 3) made little difference and did not improve model fit.

Table 5.

Factors Associated With the Number of Ways Friends and Neighbors Helped

Variable	Model 1			Model 2			Model 3
	B	SE	β	B	SE	β	B	SE	β
Constant	–0.429	0.597		–0.180	0.611		–0.233	0.633
Comorbid conditions	0.100	0.072	0.125	–0.092	0.080	–0.114	–0.092	0.080	–0.115
CES-D	0.054	0.033	0.134	0.056	0.032	0.140	0.057	0.032	0.142
Employed	0.876*	0.392	0.163	0.176	0.417	0.033	0.174	0.418	0.032
Caregiver	–0.036	0.367	–0.007	–0.076	0.358	–0.015	–0.083	0.360	–0.017
Lives alone	1.199*	0.504	0.187	1.585**	0.495	0.248	1.555**	0.505	0.243
Functional support elements	0.310*	0.155	0.144	0.295	0.150	0.137	0.294	0.150	0.137
Ways family helps	0.215**	0.069	0.242	0.242***	0.066	0.271	0.238**	0.067	0.266
Female				–0.385	0.353	–0.074	–0.269	0.498	–0.052
Research site: Uganda				2.033***	0.482	0.414	2.180**	0.657	0.444
Female × Uganda							–0.224	0.678	–0.041
R 2	0.168			0.253			0.253
Adjusted R2	0.138			0.218			0.214

Notes: Listwise N = 201.

CES-D = Center for Epidemiologic Studies Depression scale; SE = standard error.

^* p < .05

^** p < .01

^** p < .001.

Discussion and Implications

This study aimed to illuminate the role of gender in social networks and social support in two populations of adults aged 50 and older living with HIV in sub-Saharan Africa: in rural Uganda and urban and suburban South Africa. Regarding our research questions, older men and women in these samples did have different household structures and social networks: Men were more likely to be married or partnered, and women were more likely to feel close to and live with their children and grandchildren. The social network results are consistent with the body of literature (e.g., Harling et al., 2020; Mugisha et al., 2013). In both samples, older adults relied primarily on family members for support, which is also consistent with previous research (Golaz et al., 2017; Harling et al., 2020). Gender was associated with support from family members only in South Africa, where older women received family assistance at least monthly in 1.6 more ways than men did, on average, after controlling for other factors; there was no gender difference in amount of help from friends. Caregiving was not associated with receiving family support.

Overall, the data present a picture of distinct social network and support patterns among older adults living with HIV, depending on gender and environment:

In the Uganda sample, women had large families and appeared embedded in vertical family care; two-thirds were caregivers of a child or adult. Most were widowed and lived with their children and/or grandchildren. Only one woman lived alone. They usually reported feeling very close to siblings and other relatives, even if they did not see them often. Nearly all women in Uganda knew their neighbors well, and three-quarters said they “always” helped each other. Adult children and other family members helped with shopping, housework, and cooking and provided frequent emotional support. Help from friends was also common.

Men in Uganda were also embedded in family structures. Most men lived with wives or partners, and 60% lived in a household with adult children; more than half reported being a caregiver. In contrast with women, though, one-fifth lived alone. Most felt very close to other relatives, but fewer than half had frequent contact with them. Male friendship seemed quite strong. Although family members provided most instrumental support, primarily shopping and housework, friends and neighbors helped nearly as much. Friends provided regular emotional support, though one-third of men said they needed a lot more.

Women in the South Africa sample averaged 2.7 children and about three grandchildren, and nearly all women reported feeling very close to them and having frequent contact. More than half lived with children, grandchildren, or both. In contrast to women in Uganda, one-quarter lived with a partner, and one-quarter lived alone. When they needed help, one-third received it from their children, but nearly half said they received it from family members other than offspring or siblings, which could include grandchildren or mothers (MacPhail et al., 2022). Women in South Africa reported frequent family support, but this support had a negative side: Nearly half reported being upset by their family in the past month. Financial responsibilities involving the old age pension are one potential source of conflict (Button & Ncapai, 2019). Fewer than half the women reported having a functional friend.

Half the men in the South Africa sample lived with a partner, but 30% lived alone. Only about one-quarter lived with an adult child or a grandchild, and while most men felt close to children and grandchildren, a sizable share did not. Many men said they did not need physical or emotional support. Almost half had daily assistance with housework or meals, but if this was provided by a partner, they may not have perceived it as help (Schatz & Seeley, 2015). When they needed someone to talk to, men often turned to friends. Like women, they reported some negative interactions with family members.

Differences in closeness with children and grandchildren may be one reason why women received more family help, especially emotional support, than men in the South Africa sample. Labor migration and changes in family structure may play a part in these differences (Hall & Posel, 2019). In addition, declines in marriage and high rates of nonmarital childbirth have led to more female-headed households in South Africa and other sub-Saharan African countries (Odimegwu et al., 2017). Yet women often relied on other family members.

Needs were considerably greater in Uganda than in South Africa, but gender differences in needs were few. Women needed less financial help than men in South Africa but more help than men in Uganda. Women’s distinct financial advantage in South Africa over Uganda is probably due in part to South Africa’s old age pension, as well as better economic conditions overall. It is not clear why men felt more financial need than women in South Africa. In Uganda, men may have benefited from spousal income, though they were more likely than women to live alone. Nearly everyone in the Uganda sample was still working for pay in some way, but women may have earned less, or supported more family members (Mugisha et al., 2013).

According to the gendered identities and social exchange framework, women reproduce patterns of care with grandchildren, as well as expectations of reciprocity, whereas older men may feel their job is done (Golaz et al., 2017; MacPhail et al., 2022; Schatz & Seeley, 2015). We found support for this in South Africa, where men reported less connection to offspring. In Uganda, the distinction was less clear. It is possible that in environments like rural Uganda, being a caregiver is just one facet of widespread interdependence. This could help explain why caregiving was not a significant factor in the amount of support received. In Uganda, most women and men were caregivers, and most worked for money as well. Besides being driven by necessity, this situation may reflect the importance of avoiding dependence in old age, while remaining connected to children and grandchildren (Freeman, 2016; Reynolds et al., 2022).

Living alone was strongly associated with less family help but more help from friends, which suggests that individuals who live alone have built alternative support systems. Help from friends was also linked to a person’s number of functional elements (frequent contact with a child, grandchild, sibling, other relative, or friend). Greater levels of assistance may represent stronger social networks overall, or individuals who are more reliant on help. In South Africa, most older men said they did not need support, but about half turned to friends for advice or consolation at least monthly. For men who are healthy but who may lack financial resources, friendships may represent a less onerous form of interdependence than the family role of breadwinner, and living alone may be a preference.

Family members could not meet all needs. Older adults in Uganda relied on friends more than those in South Africa did. This may reflect the lack of formal services in rural areas and a country with less economic development. Although older adults in Uganda were in close contact with family members, the younger generation may have little to share (Golaz et al., 2017). Older adults in the Central Region of Uganda interviewed by Rishworth & Elliott (2022), in the absence of help from their children, built peer groups to share resources and provide emotional support. But having to rely on friends, rather than family, may leave older adults feeling dissatisfied (Brennan-Ing et al., 2022).

Strengths and Limitations

This study provides a valuable comparison of gender’s influence on social support in two distinct populations in sub-Saharan Africa. It has some limitations. As a cross-sectional study, it cannot imply causality between social factors and support from family and friends. Rural or urban residence may reflect some individuals’ preference for more or less family contact. In addition, the sample sizes by gender and research site are not large, especially for men in South Africa (n = 30), and the findings cannot be generalized to the populations in other areas. Future research should be conducted in other populations and regions of sub-Saharan Africa. The data also lack certain measures that would be helpful for the current analysis. For example, there is no measure of whether the older adults are heads of households, or what kind of support they provide to friends and family members other than caregiving. Participants were not asked about their parents, who may have provided support. Finally, this research was conducted before the COVID-19 pandemic, which presumably disrupted both formal and informal support to various degrees; further studies should examine how such disruptions have affected older men and women with HIV.

Implications

Many of the older adults living with HIV in this study supported family members and received help in return. Yet even where interdependence is strong, as in Uganda, unmet needs—physical, financial, and emotional—are considerable. A culture of interdependence cannot provide all the support required by a population with substantial health challenges, such as the high rates of health conditions and depressive symptoms among older adults living with HIV in Uganda. Given the projected growth in this population (Autenrieth et al., 2018), stronger formal supports are needed, especially in Uganda and other underserved rural areas. Community-based geriatric support, including community education, could be put in place in Uganda using the existing structure of Village Health Teams of community health volunteers (Ssensamba et al., 2022). An expansion of Uganda’s pension system, currently accessible only to adults 80 years and older, could strengthen older adults both directly and indirectly, by bolstering the investment they can make in their household members (Case & Menendez, 2007; Gelders & Athias, 2019). Pensions and other support are increasingly important in rural areas, as climate change disrupts crops and affects nutritional intake (Rishworth & Elliott, 2022). In South Africa, rural areas may have an especially pressing need for formal services, but access to such services is not consistent within urban areas either (Nxumalo et al., 2016). Both Uganda and South Africa have historically omitted older adults from the policy agenda, as significant social issues among the larger younger populations have taken precedence (Goodrick & Pelser, 2014; Rishworth & Elliott, 2022).

Conclusion

Our study affirms the gender socialization found in previous research in sub-Saharan Africa, in which women have closer ties to children and grandchildren than men do, but our findings indicate that gender roles adapt to the demands of the social and economic context. Older people with HIV, especially women, continue to provide care. Population aging demands an integrated policy response that raises the status of older people and specifically addresses their needs in the process of enabling a robust, functional society. More formal support would serve to strengthen the family and community networks on which older adults rely and which they help sustain.

Funding

This work was supported in part by HelpAge International (ROAH Uganda) and a University of New England Partnerships Career Development Grant (ROAH South Africa).

Conflict of Interest

None.