Facing the quality of life: physical illness, anxiety, and depression symptoms among people living with HIV in rural Zambia – a cross-sectional study

ABSTRACT

Widespread access to ART has not improved the quality of life (QoL) for people living with HIV (PLHIV). We used the United Nations Disability project (UNPRPD) evaluation data to examine how physical illness, anxiety, and depression shape the QoL of PLHIV in households receiving the social cash transfers safety nets in Luapula, Zambia. We explored associations between each outcome – physical illness, anxiety, depression symptoms – and age, gender, poverty, hunger and disability, using univariable and multivariable regressions. We adjusted p-values for multiple hypothesis testing with sharpened Qs. The sample comprised 1925 respondents 16–55 years old, median age 31 (IQR 22–42 years), majority women (n = 1514, 78.6%). Two-thirds (1239, 64.4%) reported having a physical illness, a third (671, 34.9%) anxiety, and nine per cent (366) depression symptoms. More HIV positive people had a disability (34.6%, 53 versus 28.3%, 502; Q = 0.033), were physically ill (72.5%, 111 versus 63.7%, 1128; Q = 0.011), and two-fold (aOR 1.97 95% CI 1.31–2.94) more likely to report depression symptoms than HIV negative peers. Food insecurity and disability among PLHIV may worsen their physical illnesses, anxiety, depression symptoms, and other QoL domains. More research on the quality of life of PLHIV in poverty is required.

KEYWORDS:

• Quality of life of people living with HIV
• anxiety
• depression
• disability
• poverty
• hunger
• cash transfers

Introduction

The AIDS treatment target of Millennium Development Goal 6 was achieved nine months ahead of schedule (UNAIDS Joint United Nations Programme on HIV/AIDS, 2015). This achievement fuelled optimism that universal access to HIV treatment by 2030 is feasible. Armed with evidence that early initiation and adherence to HIV treatment prevents illnesses and deaths, averts new infections and saves money, the 90-90-90 target was set. According to this target, by 2020, 90% of people living with HIV should know their HIV status, 90% of people with diagnosed HIV infection receive antiretroviral therapy (ART), and 90% of people receiving ART achieve viral suppression (UNAIDS Joint United Nations Programme on HIV/AIDS, 2014; UNAIDS Joint United Nations Programme on HIV/AIDS, 2016). However, people living with HIV and stakeholders advise against deprioritizing interventions that support the quality of life for people living with HIV. They assert that widespread access to ART has not addressed quality of life needs (Lazarus et al., 2016, June; Webster, 2019, June). They contend that improved quality of life enables people living with HIV to benefit from HIV services, and attain viral suppression (Lazarus et al., 2016, June; Webster, 2019, June; International HIV/AIDS Alliance, 2018; GNP+ Global Network of People Living with HIV, 2020). Others have proposed a fourth 90, sequential to, or imbedded in the 90.90.90 target (Lazarus et al., 2016, June; Webster, 2019, June; International HIV/AIDS Alliance, 2018; GNP+ Global Network of People Living with HIV, 2020; ViiV Healthcare, 2020; Guaraldi, 2019) (Lazarus, 2018). The quality of life of all people living with HIV remained central in this ongoing discussion.

The World Health Organization (WHO) defines the quality of life as people’s perceptions of their physical health, psychological and social relationships, and the wellbeing of the environment (WHO World Health Organization, 2012). Mental health difficulties, including major depressive disorders (MDD) and disability, constrain the quality of life for people living with HIV and reduce their ability to benefit from HIV services and are often used as a proxy for quality of life. A study in Malawi found a prevalence of depression of 27% among adults initiating ART (Stockton et al., 2020). In Botswana, Gupta and colleagues found in a population-based study that 30% of men and 25% of women living with HIV screened positive for depression (Gupta et al., 2010). In Uganda Kinyanda et al. and Mugisha et alobtained similar results among children who perinatally acquired HIV, youths, adults, and older people living with HIV (Kinyanda et al., 2011, 2016, December, 2017, 2019, 2020; Mugisha et al., 2016, May). Bigna et al. (2019) obtained similar results in a systematic review in sub-Saharan Africa (Bigna et al., 2019). Outside sub-Sahara Africa, people living with HIV had a two – to – four-fold higher risk of MDD than the general population (Medeiros et al., 2020). Mental health is affected by factors common in low resource settings, including physical illness, poverty, hunger and disability. Few studies have examined how these factors shape the relationships between HIV and poor mental health in Zambia and other sub-Saharan African countries. In this study, we aimed to explore whether people living with HIV had a lower quality of life than people without HIV. Our primary hypothesis was that people living with HIV experienced lower quality of life relating to physical illness, anxiety and depression symptoms than peers without HIV. Our secondary hypothesis was that older age, being female, poverty, hunger and disability are common among people living with HIV and explain our primary hypothesis results.

Methods

Data sources

The sample comprises data from households receiving the Government of Zambia Social Cash Transfer (SCT) safety net programme in four districts of Luapula province – Kawambwa, Mansa, Nchelenge and Samfya. The SCT aim at reducing extreme poverty. Only extremely poor households – living on less than USD1.25 a day – who were living in the same catchment area for at least six months meet at least one of the following conditions are eligible to receive the SCT: have a household member who is, aged 65 year or older, has a severe disability as verified by a government medical doctor, or chronically ill. Also eligible are households headed by a child below 19 years old who is not married, or by a female aged 19–64 years who is taking care of at least 3 children below 19 years old. All respondents lived in households identified as extremely poor and receiving the SCT.

We used baseline data of the impact evaluation of the United Nations Partnership for the Rights of People with Disability (UNPRPD) project in Luapula Province, Zambia. The evaluation seeks to assess the impact of social protection programmes on access and use of HIV Services in the four districts. We collected baseline data from August to September 2019. We used beneficiary payroll data for each of the four districts and sampled 90 Community Welfare Action Committees (CWACs), administrative political units, and households in two stages. In stage one, we sampled CWACs using proportional probability sampling to increase the likelihood of CWACs with a larger number of households and more concentration of services to be selected. In the second stage, we conducted a simple sampling of 25 households from each sampled CWAC.

Procedures

A written consent was obtained from the household head and household members 16 years and older to participate in the survey. Interviews were conducted confidentially by trained data collectors from outside the catchment areas in the local language. A questionnaire was administered separately to the household head and all household members 16 years and older who provided consent. The questionnaire includes questions on social demographic variables, general health, HIV testing, mental health and disability status. We drew the questions from piloted and validated instruments including from the UNICEF, University of North Carolina Carolina Population Center, Food and Agriculture Organisation supported Innocent Transfer Project Survey Tools – Zambia Child Grant Household questionnaire and the Demographic and Health Survey (DHS) and Population HIV Impact Assessment (PHIA) questions from the International Centre for Aids care and treatment programme on HIV testing and receiving test results (UNICEF, University of North Carolina Carolina Population Center, Food and Agriculture Organisation; International Centre for Aids care and treatment programme; The Demographic and Health Surveys Program, 2020). The survey instrument is attached in annex.

The University of Zambia Humanities and Social Sciences Research Ethics Committee (IRB Approval No. 2019-April-001) and the Canton of Geneva ethics committee in Geneva, Switzerland (no 2019-00500) reviewed the study protocol.

Statistical analyses

We operationalized the quality of life using the WHO definition of quality of life measured with the outcome variables being physically ill in the past two weeks, expressing anxiety, or depression symptoms all dichotomized (No, Yes). Beside age and gender, we included poverty, hunger and disability status as mediating variables, which are often used as proxy for quality of life and are common among people living with HIV. Being physically ill was operationalized with the question – Have you been sick or injured in the last two weeks? A respondent was denoted physically ill in the past two weeks if and only if they answered Yes. We assigned respondents reporting injured only to no, and those reporting physically ill and injured to yes.

Expressing anxiety or depression symptoms were derived from the Hopkins Symptoms Checklist-25 (HSCL – 25) questions. The HSCL-25 contains ten items for assessing anxiety and fifteen for depression symptoms. We used questions 2 and 10 for anxiety and 16, 21, 23 and 25 for depression. The HSCL-25 has been used widely in clinical and non-clinical settings to screen for anxiety and depression, including in Sub-Saharan Africa and among people living with HIV (Ashaba et al., 2018; Kaaya et al., 2002; Kaida et al., 2014; Velloza et al., 2017). We categorized the answers to no or yes. Outcome two was experiencing symptoms of Anxiety (no, yes) arising from combing answers to anxiety questions HSCL 2 – Have you felt very restless, like you can’t keep still? – and HSCL 10 – Have you felt very fearful? (i.e., Scared or afraid). In this study, a person was defined as experiencing symptoms of anxiety if they answered Yes, to both questions HSCL 2 and HSCL 10.

Experiencing depression was derived from the HSCL – depression questions HSCL 16 – Have you lost interest in things? (i.e., Things you usually enjoy); HSCL 21 – Have you felt very trapped or caught? (e.g., Like you are trapped in a situation you cannot get out of); HSCL 23 – Have you had a lot of trouble sleeping? And HSCL 25 – Have you felt very worthless? (i.e., Like you have no worth or value). A person was defined as experiencing depression symptoms if they answered yes to all questions HSCL 16, HSCL 21, HSCL 23, and HSCL 25.

The primary explanatory variable in our analysis was HIV testing and receiving the test results of the most recent HIV test (not tested; tested, and received an HIV negative test result; tested, and received an HIV positive test result). We excluded those who had no test result, or indeterminate test results, or who refused to disclose the test result. This question was derived from the DHS and PHIA questions on HIV testing and receiving the result. Age in years categorized as 16 - 24, 25 - 34 and 35 - 55; gender (male, female); being physically ill in the past two weeks (yes, no); experiencing hunger the past four weeks (often, rarely, no), poverty (very poor, moderately poor) and disability (yes, no). We operationalized disability with the Washington Group Short Questions (WGSQ). The WGSQ asks respondents if they have difficulties performing daily functions of seeing, hearing, walking, concentrating, self-care, and communicating. For each functional domain the level of difficulties is assigned as 1 – no difficulty, 2 – a little, 3 – a lot or 4 – can not. A person was determined to have a disability if they answered a level 3 or 4 difficulty in any functional domain (Washington Group on Disability Statistics, 2017, October; 2018). Our choice of explanatory variables was guided by our knowledge and a literature review, which suggests that age (Kinyanda et al., 2017; Liping et al., 2015), gender (Chibanda et al., 2016, February; Gupta et al., 2010; Liping et al., 2015), having a physical illness, or a disability (Abasa et al., 2020, May; Kinyanda et al., 2017) and socio economic conditions (Bernard et al., 2017, August; Kinyanda et al., 2011; Kinyanda et al., 2017; Liping et al., 2015) affect the quality of life of people living with HIV. Missing data in the variables was not imputed because fewer than 5% of data were missing overall. Data were missing for poverty (0.5%; n = 10), symptoms of anxiety (0.7%; n = 13) and depression (0.05%; n = 9).

We performed the analysis in three steps. First, we used descriptive statistics to report the characteristics of the study participants. Second, we performed univariable, and multivariable logistic regression analyses to explore associations between each outcome variable – physical illness, anxiety, and depression symptoms – and the explanatory variables. Third, we used marginal effects models to compute predicted probabilities for reporting each outcome for people testing and reporting an HIV test result, by gender and disability status, keeping all other variables at their mean values. We adjusted p-values for multiple hypotheses using Michael Anderson’s Stata code for generating the Benjamini Krieger and Yekutielie (BYK) False Discovery Rate sharpened Qs (Anderson, 2008). All p-values in this document are sharpened Qs with statistical significance set at 0.05. We clustered analysis at the CWAC level and used Stata/SE 14.1 to perform the analysis.

Results

The sample comprised 1925 respondents aged 16–55 years, with women the majority (n = 1514, 78.6%), median age 31 (interquartile range 22–42 years). Two-thirds (1265; 65.7%) reported being physically ill in the last two weeks or experiencing hunger often, i.e., Three to ten or more times in the past four weeks. Whereas 555 (28.8%) reported having a disability, 386 (20.1%) reported experiencing depression symptoms, and 153 (7.9%) reported an HIV positive test result. People reporting an HIV positive test result had a disability, were physically ill or, experienced depression symptoms than peers reporting a negative HIV test result (Table 1).

Table 1. Associations between respondents’ characteristics and their self-reported HIV test results.

Variable	HIV Negative		HIV Positive		Q value	Total
	n	%	n	%		Freq	%
Age
16 - 24	610	34.4	15	9.8		625	32.5
25 - 34	450	25.4	39	25.5		489	25.4
35 - 55	712	40.2	99	64.7		811	42.1
Total	1772	100	153	100	0.001	1925	100
Gender
Male	391	22.1	20	13.1		411	21.4
Female	1381	77.9	133	86.9		1514	78.6
Total	1772	100	153	100	0.005	1925	100
Disabled
No	1270	71.7	100	65.4		1370	71.2
Yes	502	28.3	53	34.6		555	28.8
Total	1772	100	153	100	0.033	1925	100
Hunger
No	250	14.1	24	15.7		274	14.2
Rarely	357	20.1	29	19.0		386	20.1
Often	1165	65.7	100	65.4		1265	65.7
Total	1772	100	153	100	0.558	1925	100
Poverty
Moderate	495	27.9	35	22.9		530	27.5
Very poor	1269	71.6	116	75.8		1385	71.9
Missing	8	0.5	2	1.3		10	0.5
Total	1772	100	153	100	0.130	1925	100
Physical illness
No	644	36.3	42	27.5		686	35.6
Yes	1128	63.7	111	72.5		1239	64.4
Total	1772	100	153	100	0.011	1925	100
Anxious
No	1144	64,6	97	63.4		1241	64.5
Yes	618	34,9	53	34.6		671	34.9
Missing	10	0,6	3	2.0		13	0.7
Total	1772	100	153	100	0.186	1925	100
Depressed
No	1445	81.5	105	68.6		1550	80.5
Yes	321	18.1	45	29.4		366	19.0
Missing	6	0.3	3	2.0		9	0.5
Total	1772	100	153	100	0.001	1925	100

Fisher's exact test, Q values are p values adjusted for multiple hypothesis testing.

Table 2 shows that an HIV positive test result was not associated with being physically ill, compared to a negative test result, after adjusting for age, gender, disability status, physical illness, hunger, and poverty. The odds of experiencing depression symptoms remained, however, nearly two-fold higher than among people reporting an HIV negative test result.

Table 2. Factors associated with self-reported physical illness, anxiety and depression symptoms among persons reporting testing HIV negative or positive (Adjusted*).

Factor	Odds Ratios (95% Confidence Intervals), Q values
	Physically ill	Q	Anxious	Q	Depression	Q
	n = 1915	Value	n = 1911	Value	n = 1914	Value
HIV Test result
Negative	1		1		1
Positive	1.56 (1.02 - 2.39)	0.067	0.94 (0.66 - 1.33)	0.531	1.97 (1.31 - 2.94)	0.006
Age
16 - 24	1		1		1
25 - 34	0.85 (0.67 - 1.06)	0.464	1.05 (0.85 - 1.30)	0.558	0.85 (0.64 - 1.12)	0.088
35 - 55	0.91 (0.74 - 1.12)		1.02 (0.85 - 1.24)		0.76 (0.57 - 1.01)
Gender
Male	1		1		1
Female	1.08 (0.88 - 1.33)	0.464	0.91 (0.71 - 1.16)	0.454	0.92 (0.66 - 1.29)	0.524
Physically ill
No			1		1
Yes			1.40 (1.08 - 1.83)	0.025	1.27 (0.89 - 1.80)	0.189
Disabled
No	1		1		1
Yes	1.39 (1.11 - 1.76)	0.013	1.44 (1.14 - 1.82)	0.008	1.34 (1.00 - 1.80)	0.075
Hunger
No	1		1		1
Rarely	0.86 (0.54 - 1.36)	0.51	1.44 (0.89 - 2.32)	0.012	1.24 (0.72 - 2.74)	0.274
Often	0.91 (0.63 - 1.31)		1.80 (1.19 - 2.73)		1.41 (0.72 - 2.74)
Poverty
Moderate	1		1		1
Very poor	1.05 (0.76 - 1.45)	0.55	1.56 (1.15 - 2.11)	0.012	2.14 (1.34 - 3.40)	0.006

Wald test *adjusted for age, gender, reporting physical illness the past two weeks and disability status. Q values are p-values adjusted for multiple hypothesis testing. 1 = reference.

A disability was associated with higher odds of being physically ill than no disability. Very poor compared to moderate poverty, experiencing hunger more often than not, being disabled than not, being physically ill than not were associated with higher odds of reporting anxiety symptoms. Compared to moderate poverty, more poverty was associated with a two-fold increase in reporting depression symptoms.

Table 3 shows that physical illness was predicted higher than anxiety and depression symptoms among people reporting an HIV test result. People reporting a disability, regardless of the HIV test result or gender, had a higher predicted probability of reporting physically ill and experiencing anxiety or depression symptoms than those without a disability.

Table 3. Predictive margins for reporting being sick, experiencing symptoms of anxiety and depression among persons reporting an HIV test result by disability status and gender.

Prob = probability; CI = confidence intervals
	Sick (n=1925)		Anxious (n=1912)		Depressed (n=1916)
	Prob	95% Cl	Prob	95% Cl	Prob	95% Cl
HIV negative
Not Disabled (Male)	0.60	(0.55 - 0.66)	0.33	(0.27 - 0.39)	0.17	(0.12 - 0.22)
Not Disabled (Female)	0.62	(0.59 - 0.65)	0.32	(0.28 - 0.36)	0.17	(0.14 - 0.20)
Disabled (Male)	0.68	(0.62 - 0.74)	0.43	(0.35 - 0.51)	0.22	(0.16 - 0.29)
Disabled (Female)	0.69	(0.64 - 0.73)	0.42	(0.36 - 0.48)	0,22	(0.17 - 0.27)
HIV positive
Not Disabled (Male)	0.70	(0.59 - 0.80)	0.32	(0.23 - 0.41)	0.29	(0.17 - 0.40)
Not Disabled (Female)	0.71	(0.62 - 0.80)	0.31	(0.23 - 0.39)	0.28	(0.20 - 0.37)
Disabled (Male)	0.76	(0.67 - 0.85)	0.42	(0.32 - 0.52)	0.37	(0.25 - 0.49)
Disabled (Female)	0.77	(0.69 - 0.85)	0.41	(0.31 - 0.50)	0.36	(0.27 - 0.44)

Discussion

This study explored associations between the HIV status and physical illness, anxiety, and depression symptoms, among people in extremely poor households receiving the Government of Zambia Social Cash Transfers social safety nets. We found that being HIV positive was associated with nearly two-fold more likelihood of reporting depression symptoms, but not anxiety symptoms or physical illness, than being HIV negative. A very poor self-rating on poverty status, reporting hunger often, and being disabled were associated with higher odds of reporting anxiety symptoms. A disability and being HIV positive predicted being physically ill, reporting anxiety or depression symptoms higher than being HIV negative, and having no disability.

Our finding that an HIV positive test result was significantly associated with increased odds of reporting depression symptoms contributes to the evidence asserting that depression is common and higher among people living with HIV compared to the general population (Bernard et al., 2017, August; Bigna et al., 2019; Lofgren et al., 2020; Stockton et al., 2020). In our study, a third (29.4%) of people reporting HIV positive test results reported depression symptoms compared to nearly one in five (18.1%) with an HIV negative test result. These differences persisted after adjusting for age, gender, physical illness, disability status, hunger, and poverty, suggesting that the depression symptoms were linked to HIV. The knowledge of an HIV positive status might have stimulated people living with HIV to feel trapped and unable to “get out of living with HIV” or experienced stigma, which may have been expressed as depression symptoms. A population-level study conducted in Zambia and South Africa found minor differences in both countries in anxiety and depression prevalence between the general population and people living with HIV, aware of their HIV positive status but not linked to care. Contrary to our study finding, it also found no differences between the general population and two categories of people living with HIV – those diagnosed HIV positive five years or more, and HIV positive but unaware of their HIV status (Hargreaves et al., 2018).

Our finding of no evidence of an association between an HIV positive test result and experiencing anxiety symptoms fits with the evidence that suggests that anxiety is common among people living with HIV, although not consistently higher, than among those without HIV. The prevalence of anxiety disorder in Zambia among people living with HIV was 37.8% in one study (Heuvela et al., 2013) and 10% in a population-based study (Thomas et al., 2017). It is possible, given the correlation between anxiety and depression disorders suggested by Abasa and colleagues (Abasa et al., 2020, May) that in our study some of the anxiety was accounted for in the depression symptoms, weakening the levels and associations between the anxiety variable and HIV positive test results. It is also possible that people living with HIV did not have more restlessness and fearfulness than HIV negative peers. The 2012 Stigma Index conducted in Zambia found as many as 40.5% of people living with HIV feared physical threats of violence due to their HIV positive status. However, the study did not have an HIV negative comparator group (Network of Zambian People Living with HIV, 2011). Hargreaves et al. (2018) found no major differences in anxiety and depression prevalence between the general population and people living with HIV (Hargreaves et al., 2018).

Our study also found that more HIV positive reported being physically ill than their HIV negative peers. When we controlled for age, gender, disability status, hunger and poverty, the differences disappeared. This result suggests that these factors might have worsened the health outcomes than living with HIV alone. This is consistent with the evidence asserting that the effect of HIV infection on the health-related quality of life among people living with HIV might be benign; people living with HIV have similar health, but fewer healthy years as the general population (Marcus et al., 2020; Thomas et al., 2017).

Our finding that a very poor poverty self-rating, experiencing hunger, being disabled, and physical illness factors, were associated with higher odds of reporting anxiety symptoms is novel, yet expected. This combination of associations can lead to persistent concerns about one's well-being resulting in anxiety. Studies have shown that poverty is associated with anxiety and other common mental disorders. Poverty alone, however, does not lead to anxiety (Hjelma et al., 2017; Kinyanda et al., 2011; Lund et al., 2010). The respondents were receiving cash transfers which have been shown to attenuate mental health distress (Bastagli et al., 2016; Natalia, Peterman, Seidenfeld, & Tembo, Volume 4, April 2018, pp. 225–235; Angelesa et al., 2019). With many of the stressors adjusted for in our study, testing HIV positive alone might have been insufficient to precipitate or maintain anxiety.

Having a disability has been associated with increased illness and use of health services (Bright & Kuper, 2018). In our study, the combination of being disabled and HIV positive predicted higher levels of all three outcomes – being physically ill, reporting anxiety or depression symptoms – than being HIV negative and not having a disability. Several reasons may account for this result. First, people with disability, although not always, in general have more health needs and use of health services, arising from their primary condition, than the general population (Abimanyi-Ochom et al., 2017). Second, age-related ill health is concentrated among the older age-groups, than younger groups as shown by a report from the Ministry of Community Development and Social Services of Zambia (Central Statistical Office, Ministry of Community Development and Social Services, 2018). We suspect that in our sample, the accumulation of disease and disability in older age, and being HIV positive increased probabilities of physical illnesses, experiencing anxiety, and depression symptoms (The Lancet HIV, 2019; Hanass-Hancock, Myezwa, & Carpenter, Disability and Living with HIV: Baseline from a Cohort of People on Long Term ART in South Africa, 2015; Hanass-Hancock et al., 2020).

Our study has several limitations. It is cross-sectional, based on self-reports and unable to determine causation. The variables we considered may not reflect the presence and duration of HIV, anxiety and depression, which might bias our results. Our sample was from respondents living in extremely poor households receiving social cash transfers. We could not generalize our results to people not receiving social cash transfers or outside our study area. However, our study enabled us to understand associations between physical illness, anxiety, depression symptoms and several key stressors: extreme poverty, hunger, disability and being HIV positive, typical features of social protection including cash transfer programmes in sub-Sahara Africa.

Conclusion

Food insecurity and disability among people living with HIV may worsen their physical illnesses, anxiety, and depression symptoms, and other quality of life domains. People living in poverty might also benefit from health-related quality of life improvement. More research on factors to improve the quality of life of people living with HIV in poverty is required.

Geolocation information.

The GPS coordinates of Luapula Province, Zambia are DD COORDINATES – 11.0 29.0 DMS Coordinate – 11°00'0.00” S 29°00'0.00” E Geohash coordinates, kwh97 UTM Coordinates 35L 718529.11253277 8783292.5914355.

Additional files

Impact of Social Protection on access and use of HIV services – Household Questionnaire.

Acknowledgements

Disclaimer: However, information in this article are the views of authors and not of UNAIDS, ILO, United Nations Partnership for the Rights of People with Disabilities (UNPRPD) or the institutions listed.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Because the data set contains unique identifiers and participants of this study did not agree for their data to be shared publicly, supporting data is not available.

Additional information

Funding

This work was supported by funding from the United Kingdom Research in Innovation (UKRI) Global Challenges Research Fund (GCRF) Accelerating Achievement for Africa’s Adolescents Hub. Additional support was received from the Swiss National Science Foundation (grant no 163878). The United Nations Joint Programme on HIV/AIDS (UNAIDS), the International Labour Organisation (ILO) and the Ministries of Health, Community Development and Social Services in Zambia provided in-kind support.