Human immunodeficiency virus (HIV) is an incurable virus that attacks the immune system and affects around 39 million people worldwide. Once diagnosed, HIV can be treated with antiretroviral therapy (ART) to limit its effects and stop it spreading to other people. HIV rates vary across the world, but the African country of Eswatini has the highest prevalence with more than one in four (27%) people classed as HIV-positive.

Until 2015, people living with HIV were typically only treated with ART once their immune system weakened. Recent studies found that starting treatment earlier enhances the positive effects of ART. This caused the World Health Organization (WHO) to change their guidelines and advise people living with HIV to begin ART as soon as they are diagnosed. While antiretroviral drugs are usually provided to patients free of charge, accessing care can be expensive for patients because of high transport costs or lost time from income-generating activities. This means starting treatment earlier and, thus, having more frequent healthcare visits, may result in a greater cost to the patient. The economic impact of this change is unclear, and for patients living in poverty, these added costs can affect their decision on whether to continue treatment.

Steinert et al. interviewed 3,019 HIV-patients from 14 health facilities in Eswatini who began treatment with ART either immediately after diagnosis or after their immune system became suppressed. Patients were asked about their time spent being active to generate income, employment status, monthly household expenditures, and household living standards. On average, beginning ART earlier appears to have had no large negative effects on the economic wellbeing of patients. The same results were found for patient groups defined by sex, education, age, and time spent taking ART.

These findings suggest that starting ART for HIV as soon as possible offers medical benefits and seems to have no large economic consequences for patients in the short term, even for poorer communities. This adds weight to the WHO advice on HIV treatment and supports the need to continue to deliver effective treatments to countries like Eswatini that have a high rate of HIV infection.

Recent trials have pointed to substantial health benefits of immediate antiretroviral therapy (ART) initiation for all HIV-positive patients compared to initiating ART based on a CD4-cell count threshold. Benefits include reduced HIV-related mortality and morbidity and decreased transmission risk to HIV-negative sexual partners (Danel et al., 2015; Cohen et al., 2016; Lundgren et al., 2015; Hayes et al., 2019; Ford et al., 2018). In line with this epidemiological evidence, the World Health Organization (WHO) has updated its consolidated guidelines on the use of antiretrovirals in 2015, now advocating for immediate ART initiation (or ‘universal test and treat’) for all HIV-positive adults, adolescents, and children (WHO, 2019).

In view of these major changes in ART provision, it is crucial for health policy makers to understand the implications of immediate ART initiation for HIV patients’ economic outcomes. At high CD4-count levels, we would expect the majority of patients to be relatively healthy, and thus have productivity levels and out-of-pocket health expenditures that are similar to those of HIV-negative patients (Thirumurthy et al., 2013). While ART may still improve economic welfare among these patients through an improvement in health status, it may also decrease these patients’ economic welfare through, for example, the side effects of antiretroviral drugs, increased frequency of (ART) clinic visits or stigma from taking ART (daCosta DiBonaventura et al., 2012; Unge et al., 2008). The economic consequences of early ART initiation for this specific patient group are therefore unclear and have to date not been investigated experimentally.

Previous studies have assessed labour market outcomes and overall financial wellbeing of patients on ART, relative to patients not yet on ART. Of these, several studies have highlighted beneficial economic impacts of ART initiation, which are primarily based on the positive labour market effects of improved health. Accordingly, empirical evidence has pointed to higher work performance and productivity, (Bor et al., 2012; Larson et al., 2008; Beard et al., 2009; Thirumurthy et al., 2008) lower absenteeism, (Larson et al., 2008) increases in savings rates, (Baranov and Kohler, 2018) as well as increased educational expenditures and attainment (Baranov and Kohler, 2018; Lucas et al., 2019) following ART initiation. Conversely, other studies have documented detrimental economic effects of ART initiation (even under universal access to ART schemes), largely driven by three suggested mechanisms: first, by increased patient-borne healthcare expenditures associated with travel to ART clinics, clinic and hospital fees, and income foregone; (Rosen et al., 2007; Chimbindi et al., 2015; Leive and Xu, 2008) second, by elevated levels of food insecurity due to a treatment-induced increase in appetite and fewer financial resources to absorb the higher food expenditures; (Patenaude et al., 2018) and third, by reduced productivity as a result of short-term adverse and toxic effects linked to antiretroviral drugs (Danel et al., 2015; Rosen et al., 2007). However, these previous studies provide only little insights on the anticipated economic effects of immediate ART initiation because they are based on outdated treatment guidelines, thus comparing HIV-patients above and below a certain CD4 cell count level (e.g. 500 cells/mm³) that determines ART eligibility. Given that HIV-patients who are not yet on ART but have a relatively low CD4 count may be more susceptible to opportunistic infections and adverse events than those with higher CD4 counts, this comparison group is inadequate for assessing the economic consequences of the current WHO-endorsed ART initiation strategy that is independent of patients’ CD4 counts.

To decide whether and how much governments and international organisations should invest in scaling up immediate ART initiation for all HIV-patients, it is crucial to understand the impact of immediate ART initiation not only on health but also on HIV patients’ economic outcomes. This is the first randomised trial aimed at answering this question. Specifically, we conducted a stepped-wedge cluster-randomised controlled trial of the ‘Early Initiation of ART for All’ (EAAA) intervention for HIV-patients in Eswatini to test the causal impact of immediate ART initiation on a range of economic outcomes, including patients’ time use, employment, household expenditures, and household living standards.

We present the first causal evaluation of the effect of immediate ART for all HIV patients on wider economic outcomes. Based on our primary results and several robustness checks, we are able to conclude that large harmful effects are very unlikely. More specifically, we found that neither patients’ time use nor their employment status and living standards were positively or negatively affected by the EAAA intervention. Although we found a reduction in monthly household expenditures among patients in the EAAA group, the magnitude was small in size (−126.17 SZL, corresponding to 3% of the average monthly household expenditures in Eswatini) (The World Bank, 2019) and not statistically significant. Lastly, in machine-learning-supported heterogeneity analyses, we also did not find any patient subgroup for which the EAAA intervention either significantly improved or deteriorated individual- and household-level economic outcomes. Two previous publications based on the same trial have assessed how early ART initiation affected patients’ health, revealing a 6% higher retention in care rate in the EAAA group but no significant differences with regard to all-cause, disease-related, and HIV-related mortality between the EAAA and the standard of care group (Chao et al., 2020; Perriat et al., 2018). While we were unable to link responses from this survey to patients’ clinical data, we may still infer that more substantial health impacts would have been necessary to significantly affect patients’ economic welfare. It is also possible that both the health and economic benefits of early ART initiation only materialise after a longer follow-up time, beyond the 36 months observation period covered in this trial (May et al., 2016). A potential alternative explanation for the absence of strong and beneficial treatment effects could relate to the broader socioeconomic conditions of the study region. Hence, if income generation opportunities are generally constrained due to given economic circumstances, people living with HIV may be unable to find work, irrespective of whether they are healthy or not. If patients’ health status does not substantially impact their earning potential, other welfare indicators such as household expenditures and living standards are also unlikely to change. However, we partly alleviate this problem by adopting a broad definition of employment by including informal and short-term work and should therefore be able to capture even small changes in participants’ income generation activities.

In contrast to several prior studies (Beard et al., 2009; Thirumurthy et al., 2008; Habyarimana et al., 2010; Baranov and Kohler, 2018), our findings did not exhibit any substantial detrimental financial and economic consequences of ART initiation. At the very least, our results suggest that ART-related adverse events were not substantial enough to provoke significant drops in patients’ productivity levels (Nansseu and Bigna, 2017). We therefore add important new empirical evidence from the perspective of patients’ economic wellbeing, which – given that EAAA does not appear to have large detrimental effects on patients’ economic outcomes – support the 2015 WHO recommendation to offer immediate ART initiation to all HIV-patients.

Our study has several key strengths. First, we examined a comprehensive set of outcome variables and are thus able to gain nuanced insights into participants’ overall economic situation. Although the different outcomes are likely correlated, time use and employment are patient-level variables, whereas expenditures and living standards are captured at the household level. The latter two variables could thus be differently affected by the EAAA intervention depending on whether the patient is the household’s main breadwinner or not. Household savings could have been another possible welfare-related aspect to assess. However, the general savings rate in Eswatini is low (World Bank Group, 2017) and savings are often mainly used to smooth consumption, and thus likely highly correlated with overall household expenditures. Second, we have collected very detailed information on patients’ time use. Time use is a measure that is presumably highly sensitive to potential short-term changes in economic productivity and, given that we asked about the previous 24 hr, less prone to measurement error or bias from a long recall period. In view of the precise null effects for the time use outcome, we can more confidently conclude that immediate ART initiation had no harmful effects on patients’ overall productivity levels. Third, and arguably most importantly, this is the first randomised study - and thus the first study to allow for causal inference under no untestable assumptions – of the impact of immediate ART initiation on indicators of patients’ economic outcomes.

This study has six main limitations. First, biological data on patients’ CD4 count levels and viral loads were not collected. It was thus not possible to assess whether the effects of EAAA on patients’ economic outcomes were different among those patients who had a CD4 count close to the treatment threshold at the time of ART initiation. Second, participant recruitment was implemented within healthcare facilities and it is therefore possible that patients who generally attend healthcare services more regularly and reliably were overrepresented in the study sample. Third, participants were not followed-up on longitudinally, which implies that for each individual, we either had a measurement of the pre- or the post-intervention phase (but never for both). Our effect estimates are based on the comparison of patients in the standard of care phase with patients in the EAAA phase, and would turn invalid if there was significant imbalance in baseline characteristics between these two groups. However, this is unlikely in view of the sufficiently large sample size and the random selection of interview dates for each facility. Fourth, data on household expenditures and household living standards were only collected from a random subsample of 50% of patients. Given the wide confidence interval of the effect estimates for household expenditures , it is possible that we would have been able to find a significant effect for this outcome of a size that would still be meaningful to health policy makers if we had had a larger sample size. Fifth, data was based on patients’ self-report. Especially with regard to household expenditures and time use, this limitation is likely to have led to some degree of measurement error due to recall problems (Filmer and Pritchett, 2001; Sahn and Stifel, 2000). In addition, while monthly expenses were summarised into 20 distinct expenditure categories to reduce interview length and cost, this may have led to further measurement imprecisions, for instance through adding up expenses for numerous individual food items into an overall category of ‘total shopping for food and groceries’. Yet, we expect that these measurement errors and reporting biases occurred – on average – to an equal degree in the EAAA and standard of care group and are therefore unlikely to systematically bias our point estimates of the causal intervention effect. Lastly, the employment rate in our study sample diverged from the national employment rate during the same period. This discrepancy could be explained by (i) the composition of our sample, which consisted of 75% female patients and is therefore not representative for the population as a whole, (ii) the temporal disaggregation into tertials, which might reflect some seasonal fluctuations in our data, and (iii) the lack of regional labour force data for the general population in the Hhohho region, rather than the aggregated national data that we have used as a reference.

This study provides the first causal evidence on the effect of immediate ART initiation on individual- and household-level economic outcomes. EAAA is unlikely to have detectable, harmful economic repercussions for HIV patients in Eswatini. This is an important finding for health policy making in that it buttresses the WHO recommendation to discard eligibility thresholds for ART from the perspective of patients’ economic wellbeing – a perspective that is often ignored in the setting of clinical recommendations, yet important to those who are directly affected by these recommendations (Govindasamy et al., 2012; Fox et al., 2010).

The Maximising ART for Better Health and Zero New HIV Infections (MaxART) trial (Walsh et al., 2017) (NCT02909218) and the economic outcome analysis presented in here (NCT03789448) were pre-registered on ClinicalTrials.gov.

Study setting

Request a detailed protocol

The study was implemented in North-Western Eswatini (formerly ‘Swaziland’). 27.0% of the general population in Eswatini are HIV-positive; the highest HIV prevalence worldwide (Government of the Kingdom of Eswatini, 2017). The trial enrolled 14 government-managed health facilities located in the Hhohho region (see Figure 4). At the study’s outset in 2014, all health facilities provided comprehensive HIV care and treatment according to the national adult HIV treatment guidelines effective at the time, thus initiating ART according to prescribed CD4 count levels. According to the Annual HIV Program Report of 2014 (Kingdom of Swaziland, Ministry of Health, 2014), almost 60% of HIV-patients in the Hhohho region had been initiated on ART in the year prior to the trial roll-out.

Figure 4

Download asset Open asset

All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 2-4 and all supplementary Figures (Figures S1-S9).

1. 1. Athey S
   2. Tibshirani J
   3. Wager S

   (2019) Generalized random forests

   The Annals of Statistics 47:1148–1178.

   https://doi.org/10.1214/18-AOS1709

   • Google Scholar
2. Preprint

   1. Athey S
   2. Imbens G

   (2016) The econometrics of randomized experiments

   arXiv.

   http://arxiv.org/abs/1607.00698

   • Google Scholar
3. Preprint

   1. Athey S
   2. Wager S

   (2019) Estimating treatment effects with causal forests: an application

   arXiv.

   http://arxiv.org/abs/1902.07409

   • Google Scholar
4. 1. Azur MJ
   2. Stuart EA
   3. Frangakis C
   4. Leaf PJ

   (2011) Multiple imputation by chained equations: what is it and how does it work?

   International Journal of Methods in Psychiatric Research 20:40–49.

   https://doi.org/10.1002/mpr.329

   • PubMed
   • Google Scholar
5. 1. Baranov V
   2. Kohler H-P

   (2018) The impact of AIDS treatment on savings and human capital investment in Malawi

   American Economic Journal: Applied Economics 10:266–306.

   https://doi.org/10.1257/app.20150369

   • Google Scholar
6. 1. Beard J
   2. Feeley F
   3. Rosen S

   (2009) Economic and quality of life outcomes of antiretroviral therapy for HIV/AIDS in developing countries: a systematic literature review

   AIDS Care 21:1343–1356.

   https://doi.org/10.1080/09540120902889926

   • PubMed
   • Google Scholar
7. 1. Bor J
   2. Tanser F
   3. Newell M-L
   4. Bärnighausen T

   (2012) Nearly full employment recovery among south african HIV patients on antiretroviral therapy: evidence from A large population cohort

   Health Aff (Millwood) 31:0407.

   https://doi.org/10.1377/hlthaff.2012.0407

   • Google Scholar
8. 1. Chao A
   2. Spiegelman D
   3. Khan S
   4. Walsh F
   5. Mazibuko S
   6. Pasipamire M
   7. Chai B
   8. Reis R
   9. Mlambo K
   10. Delva W
   11. Khumalo G
   12. Zwane M
   13. Fleming Y
   14. Mafara E
   15. Hettema A
   16. Lejeune C
   17. Bärnighausen T
   18. Okello V

   (2020) Mortality under early access to antiretroviral therapy vs. Eswatini's national standard of care: the MaxART clustered randomized stepped-wedge trial

   HIV Medicine 21:429–440.

   https://doi.org/10.1111/hiv.12876

   • PubMed
   • Google Scholar
9. 1. Chimbindi N
   2. Bor J
   3. Newell ML
   4. Tanser F
   5. Baltussen R
   6. Hontelez J
   7. de Vlas SJ
   8. Lurie M
   9. Pillay D
   10. Bärnighausen T

   (2015) Time and money: the true costs of health care utilization for patients receiving "Free" HIV/Tuberculosis Care and Treatment in Rural KwaZulu-Natal

   Journal of Acquired Immune Deficiency Syndromes 70:e52–e60.

   https://doi.org/10.1097/QAI.0000000000000728

   • PubMed
   • Google Scholar
10. 1. Cohen MS
    2. Chen YQ
    3. McCauley M
    4. Gamble T
    5. Hosseinipour MC
    6. Kumarasamy N
    7. Hakim JG
    8. Kumwenda J
    9. Grinsztejn B
    10. Pilotto JH
    11. Godbole SV
    12. Chariyalertsak S
    13. Santos BR
    14. Mayer KH
    15. Hoffman IF
    16. Eshleman SH
    17. Piwowar-Manning E
    18. Cottle L
    19. Zhang XC
    20. Makhema J
    21. Mills LA
    22. Panchia R
    23. Faesen S
    24. Eron J
    25. Gallant J
    26. Havlir D
    27. Swindells S
    28. Elharrar V
    29. Burns D
    30. Taha TE
    31. Nielsen-Saines K
    32. Celentano DD
    33. Essex M
    34. Hudelson SE
    35. Redd AD
    36. Fleming TR
    37. HPTN 052 Study Team

    (2016) Antiretroviral therapy for the prevention of HIV-1 transmission

    New England Journal of Medicine 375:830–839.

    https://doi.org/10.1056/NEJMoa1600693

    • PubMed
    • Google Scholar
11. 1. Crump RK
    2. Hotz VJ
    3. Imbens GW
    4. Mitnik OA

    (2008) Nonparametric tests for treatment effect heterogeneity

    Review of Economics and Statistics 90:389–405.

    https://doi.org/10.1162/rest.90.3.389

    • Google Scholar
12. 1. daCosta DiBonaventura M
    2. Gupta S
    3. Cho M
    4. Mrus J

    (2012) The association of HIV/AIDS treatment side effects with health status, work productivity, and resource use

    AIDS Care 24:744–755.

    https://doi.org/10.1080/09540121.2011.630363

    • PubMed
    • Google Scholar
13. 1. Danel C
    2. Moh R
    3. Gabillard D
    4. Badje A
    5. Le Carrou J
    6. Ouassa T
    7. Ouattara E
    8. Anzian A
    9. Ntakpé JB
    10. Minga A
    11. Kouame GM
    12. Bouhoussou F
    13. Emieme A
    14. Kouamé A
    15. Inwoley A
    16. Toni TD
    17. Ahiboh H
    18. Kabran M
    19. Rabe C
    20. Sidibé B
    21. Nzunetu G
    22. Konan R
    23. Gnokoro J
    24. Gouesse P
    25. Messou E
    26. Dohoun L
    27. Kamagate S
    28. Yao A
    29. Amon S
    30. Kouame AB
    31. Koua A
    32. Kouamé E
    33. Ndri Y
    34. Ba-Gomis O
    35. Daligou M
    36. Ackoundzé S
    37. Hawerlander D
    38. Ani A
    39. Dembélé F
    40. Koné F
    41. Guéhi C
    42. Kanga C
    43. Koule S
    44. Séri J
    45. Oyebi M
    46. Mbakop N
    47. Makaila O
    48. Babatunde C
    49. Babatounde N
    50. Bleoué G
    51. Tchoutedjem M
    52. Kouadio AC
    53. Sena G
    54. Yededji SY
    55. Assi R
    56. Bakayoko A
    57. Mahassadi A
    58. Attia A
    59. Oussou A
    60. Mobio M
    61. Bamba D
    62. Koman M
    63. Horo A
    64. Deschamps N
    65. Chenal H
    66. Sassan-Morokro M
    67. Konate S
    68. Aka K
    69. Aoussi E
    70. Journot V
    71. Nchot C
    72. Karcher S
    73. Chaix ML
    74. Rouzioux C
    75. Sow PS
    76. Perronne C
    77. Girard PM
    78. Menan H
    79. Bissagnene E
    80. Kadio A
    81. Ettiegne-Traore V
    82. Moh-Semdé C
    83. Kouame A
    84. Massumbuko JM
    85. Chêne G
    86. Dosso M
    87. Domoua SK
    88. N'Dri-Yoman T
    89. Salamon R
    90. Eholié SP
    91. Anglaret X
    92. TEMPRANO ANRS 12136 Study Group

    (2015) A trial of early antiretrovirals and isoniazid preventive therapy in africa

    New England Journal of Medicine 373:808–822.

    https://doi.org/10.1056/NEJMoa1507198

    • PubMed
    • Google Scholar
14. 1. Filmer D
    2. Pritchett LH

    (2001) Estimating wealth effects without expenditure Data-or tears: an application to educational enrollments in states of India

    Demography 38:115–132.

    https://doi.org/10.1353/dem.2001.0003

    • PubMed
    • Google Scholar
15. 1. Ford N
    2. Migone C
    3. Calmy A
    4. Kerschberger B
    5. Kanters S
    6. Nsanzimana S
    7. Mills EJ
    8. Meintjes G
    9. Vitoria M
    10. Doherty M
    11. Shubber Z

    (2018) Benefits and risks of rapid initiation of antiretroviral therapy

    AIDS 32:17–23.

    https://doi.org/10.1097/QAD.0000000000001671

    • PubMed
    • Google Scholar
16. 1. Fox MP
    2. Mazimba A
    3. Seidenberg P
    4. Crooks D
    5. Sikateyo B
    6. Rosen S

    (2010) Barriers to initiation of antiretroviral treatment in rural and urban Areas of Zambia: a cross-sectional study of cost, stigma, and perceptions about ART

    Journal of the International AIDS Society 13:8.

    https://doi.org/10.1186/1758-2652-13-8

    • PubMed
    • Google Scholar
17. 1. Geldsetzer P
    2. Fink G
    3. Vaikath M
    4. Bärnighausen T

    (2018) Sampling for patient exit interviews: assessment of methods using mathematical derivation and computer simulations

    Health Services Research 53:256–272.

    https://doi.org/10.1111/1475-6773.12611

    • PubMed
    • Google Scholar
18. Report

    1. Government of the Kingdom of Eswatini

    (2017)

    Swaziland HIV Incidence Measurement Survey 2: A Population-Based HIV Impact Assessment

    Mbabane ICAP 2017 Published online.

    • Google Scholar
19. 1. Govindasamy D
    2. Ford N
    3. Kranzer K

    (2012) Risk factors, barriers and facilitators for linkage to antiretroviral therapy care: a systematic review

    AIDS 26:2059–2067.

    https://doi.org/10.1097/QAD.0b013e3283578b9b

    • PubMed
    • Google Scholar
20. 1. Habyarimana J
    2. Mbakile B
    3. Pop-Eleches C

    (2010) The impact of HIV/AIDS and ARV treatment on worker absenteeism: implications for african firms

    The Journal of Human Resources 45:809–839.

    https://doi.org/10.1353/jhr.2010.0032

    • Google Scholar
21. 1. Hayes RJ
    2. Donnell D
    3. Floyd S
    4. Mandla N
    5. Bwalya J
    6. Sabapathy K
    7. Yang B
    8. Phiri M
    9. Schaap A
    10. Eshleman SH
    11. Piwowar-Manning E
    12. Kosloff B
    13. James A
    14. Skalland T
    15. Wilson E
    16. Emel L
    17. Macleod D
    18. Dunbar R
    19. Simwinga M
    20. Makola N
    21. Bond V
    22. Hoddinott G
    23. Moore A
    24. Griffith S
    25. Deshmane Sista N
    26. Vermund SH
    27. El-Sadr W
    28. Burns DN
    29. Hargreaves JR
    30. Hauck K
    31. Fraser C
    32. Shanaube K
    33. Bock P
    34. Beyers N
    35. Ayles H
    36. Fidler S
    37. HPTN 071 (PopART) Study Team

    (2019) Effect of universal testing and treatment on HIV incidence - HPTN 071 (PopART)

    New England Journal of Medicine 381:207–218.

    https://doi.org/10.1056/NEJMoa1814556

    • PubMed
    • Google Scholar
22. 1. Hussey MA
    2. Hughes JP

    (2007) Design and analysis of stepped wedge cluster randomized trials

    Contemporary Clinical Trials 28:182–191.

    https://doi.org/10.1016/j.cct.2006.05.007

    • PubMed
    • Google Scholar
23. Report

    1. Kingdom of Swaziland, Ministry of Health

    (2014)

    Annual HIV Program Report

    HTC, PMTCT, and ART.

    • Google Scholar
24. 1. Larson BA
    2. Fox MP
    3. Rosen S
    4. Bii M
    5. Sigei C
    6. Shaffer D
    7. Sawe F
    8. Wasunna M
    9. Simon JL

    (2008) Early effects of antiretroviral therapy on work performance: preliminary results from a cohort study of kenyan agricultural workers

    AIDS 22:421–425.

    https://doi.org/10.1097/QAD.0b013e3282f3cc0c

    • PubMed
    • Google Scholar
25. 1. Lee M

    (2009) Non-parametric tests for distributional treatment effect for randomly censored responses

    Journal of the Royal Statistical Society: Series B 71:243–264.

    https://doi.org/10.1111/j.1467-9868.2008.00683.x

    • Google Scholar
26. 1. Leive A
    2. Xu K

    (2008) Coping with out-of-pocket health payments: empirical evidence from 15 African countries

    Bulletin of the World Health Organization 86:849–856.

    https://doi.org/10.2471/BLT.07.049403

    • PubMed
    • Google Scholar
27. 1. Lucas AM
    2. Chidothe M
    3. Wilson NL

    (2019) Effects of adult health interventions at scale on children’s schooling: Evidence from antiretroviral therapy in Zambia

    Economics of Education Review 72:107–120.

    https://doi.org/10.1016/j.econedurev.2019.04.006

    • Google Scholar
28. 1. Lundgren JD
    2. Babiker AG
    3. Gordin F
    4. Emery S
    5. Grund B
    6. Sharma S
    7. Avihingsanon A
    8. Cooper DA
    9. Fätkenheuer G
    10. Llibre JM
    11. Molina JM
    12. Munderi P
    13. Schechter M
    14. Wood R
    15. Klingman KL
    16. Collins S
    17. Lane HC
    18. Phillips AN
    19. Neaton JD
    20. INSIGHT START Study Group

    (2015) Initiation of antiretroviral therapy in early asymptomatic HIV infection

    New England Journal of Medicine 373:795–807.

    https://doi.org/10.1056/NEJMoa1506816

    • PubMed
    • Google Scholar
29. 1. May MT
    2. Vehreschild JJ
    3. Trickey A
    4. Obel N
    5. Reiss P
    6. Bonnet F
    7. Mary-Krause M
    8. Samji H
    9. Cavassini M
    10. Gill MJ
    11. Shepherd LC
    12. Crane HM
    13. d'Arminio Monforte A
    14. Burkholder GA
    15. Johnson MM
    16. Sobrino-Vegas P
    17. Domingo P
    18. Zangerle R
    19. Justice AC
    20. Sterling TR
    21. Miró JM
    22. Sterne JAC
    23. Boulle A
    24. Stephan C
    25. Miro JM
    26. Cavassini M
    27. Chêne G
    28. Costagliola D
    29. Dabis F
    30. Monforte AD
    31. Del Amo J
    32. Van Sighem A
    33. Fätkenheuer G
    34. Gill J
    35. Guest J
    36. Haerry DH
    37. Hogg R
    38. Justice A
    39. Shepherd L
    40. Obel N
    41. Crane H
    42. Smith C
    43. Reiss P
    44. Saag M
    45. Sterling T
    46. Teira R
    47. Williams M
    48. Zangerle R
    49. Sterne J
    50. May M
    51. Ingle S
    52. Trickey A
    53. Antiretroviral Therapy Cohort Collaboration (ART-CC)

    (2016) Mortality according to CD4 count at start of combination antiretroviral therapy among HIV-infected patients followed for up to 15 years after start of treatment: collaborative cohort study

    Clinical Infectious Diseases 62:1571–1577.

    https://doi.org/10.1093/cid/ciw183

    • PubMed
    • Google Scholar
30. Report

    1. Ministry of Health, Kingdom of Swaziland

    (2015) Swaziland Integrated HIV Managment Guidelines

    WHO.

    https://hivstar.lshtm.ac.uk/files/2017/11/Swaziland-integrated-HIV-management-guidelines-2015.pdf

    • Google Scholar
31. 1. Nansseu JR
    2. Bigna JJ

    (2017) Antiretroviral therapy related adverse effects: can sub-Saharan africa cope with the new "test and treat" policy of the World Health Organization?

    Infectious Diseases of Poverty 6:24.

    https://doi.org/10.1186/s40249-017-0240-3

    • PubMed
    • Google Scholar
32. 1. Patenaude BN
    2. Chimbindi N
    3. Pillay D
    4. Bärnighausen T

    (2018) The impact of ART initiation on household food security over time

    Social Science & Medicine 198:175–184.

    https://doi.org/10.1016/j.socscimed.2017.11.036

    • PubMed
    • Google Scholar
33. 1. Perriat D
    2. Balzer L
    3. Hayes R
    4. Lockman S
    5. Walsh F
    6. Ayles H
    7. Floyd S
    8. Havlir D
    9. Kamya M
    10. Lebelonyane R
    11. Mills LA
    12. Okello V
    13. Petersen M
    14. Pillay D
    15. Sabapathy K
    16. Wirth K
    17. Orne-Gliemann J
    18. Dabis F
    19. the Universal Test and Treat Trials Consortium (UT3C)

    (2018) Comparative assessment of five trials of universal HIV testing and treatment in sub-Saharan africa

    Journal of the International AIDS Society 21:e25048.

    https://doi.org/10.1002/jia2.25048

    • PubMed
    • Google Scholar
34. 1. Rosen S
    2. Ketlhapile M
    3. Sanne I
    4. DeSilva MB

    (2007)

    Cost to patients of obtaining treatment for HIV/AIDS in South Africa

    South African Medical Journal = Suid-Afrikaanse Tydskrif Vir Geneeskunde 97:524–529.

    • PubMed
    • Google Scholar
35. 1. Royston P
    2. White I

    (2011) Multiple Imputation by Chained Equations (MICE): Implementation in Stata

    Journal of Statistical Software 45:i04.

    https://doi.org/10.18637/jss.v045.i04

    • Google Scholar
36. 1. Sahn DE
    2. Stifel DC

    (2000) Poverty comparisons over time and across countries in africa

    World Development 28:2123–2155.

    https://doi.org/10.1016/S0305-750X(00)00075-9

    • Google Scholar
37. 1. Sahn DE
    2. Stifel D

    (2003) Exploring alternative measures of welfare in the absence of expenditure data

    Review of Income and Wealth 49:463–489.

    https://doi.org/10.1111/j.0034-6586.2003.00100.x

    • Google Scholar
38. Report

    1. The World Bank

    (2019)

    Eswatini Household Income and Expenditure Survey

    2016/17 Final Report.

    • Google Scholar
39. 1. Thirumurthy H
    2. Zivin JG
    3. Goldstein M

    (2008) The economic impact of AIDS treatment: labor supply in western Kenya

    Journal of Human Resources 43:511–552.

    https://doi.org/10.1353/jhr.2008.0009

    • PubMed
    • Google Scholar
40. 1. Thirumurthy H
    2. Chamie G
    3. Jain V
    4. Kabami J
    5. Kwarisiima D
    6. Clark TD
    7. Geng E
    8. Petersen ML
    9. Charlebois ED
    10. Kamya MR
    11. Havlir DV
    12. SEARCH Collaboration

    (2013) Improved employment and education outcomes in households of HIV-infected adults with high CD4 cell counts: evidence from a community health campaign in Uganda

    AIDS 27:627–634.

    https://doi.org/10.1097/QAD.0b013e32835c54d8

    • PubMed
    • Google Scholar
41. 1. Thompson JA
    2. Fielding KL
    3. Davey C
    4. Aiken AM
    5. Hargreaves JR
    6. Hayes RJ

    (2017) Bias and inference from misspecified mixed-effect models in stepped wedge trial analysis

    Statistics in Medicine 36:3670–3682.

    https://doi.org/10.1002/sim.7348

    • PubMed
    • Google Scholar
42. Website

    1. Thompson J

    (2019) SWPERMUTE: stata module to compute permutation tests for Stepped-Wedge Cluster-Randomised trials

    Boston College Department of Economics. Accessed October 30, 2019.

    https://ideas.repec.org/c/boc/bocode/s458426.html
43. 1. Unge C
    2. Johansson A
    3. Zachariah R
    4. Some D
    5. Van Engelgem I
    6. Ekstrom AM

    (2008) Reasons for unsatisfactory acceptance of antiretroviral treatment in the urban Kibera slum, Kenya

    AIDS Care 20:146–149.

    https://doi.org/10.1080/09540120701513677

    • PubMed
    • Google Scholar
44. 1. Wager S
    2. Athey S

    (2018) Estimation and inference of heterogeneous treatment effects using random forests

    Journal of the American Statistical Association 113:1228–1242.

    https://doi.org/10.1080/01621459.2017.1319839

    • Google Scholar
45. 1. Walsh FJ
    2. Bärnighausen T
    3. Delva W
    4. Fleming Y
    5. Khumalo G
    6. Lejeune CL
    7. Mazibuko S
    8. Mlambo CK
    9. Reis R
    10. Spiegelman D
    11. Zwane M
    12. Okello V

    (2017) Impact of early initiation versus national standard of care of antiretroviral therapy in Swaziland's public sector health system: study protocol for a stepped-wedge randomized trial

    Trials 18:383.

    https://doi.org/10.1186/s13063-017-2128-8

    • PubMed
    • Google Scholar
46. Website

    1. WHO

    (2019) Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection

    WHO. Accessed April 1, 2019.

    http://www.who.int/hiv/pub/arv/arv-2016/en/
47. Website

    1. World Bank Group

    (2017) The global findex database measuring financial inclusion and the fintech revolution 2017

    Accessed September 11, 2020.

    https://globalfindex.worldbank.org/
48. 1. Yu LM
    2. Burton A
    3. Rivero-Arias O

    (2007) Evaluation of software for multiple imputation of semi-continuous data

    Statistical Methods in Medical Research 16:243–258.

    https://doi.org/10.1177/0962280206074464

    • PubMed
    • Google Scholar

Author details

1. Janina Isabel Steinert

   Technical University of Munich, Munich, Germany

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   For correspondence

   janina.steinert@tum.de

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7120-0075

2. Shaukat Khan

   Clinton Health Acccess Initiative, Boston, United States

   Contribution

   Data curation, Project administration

   Competing interests

   No competing interests declared

3. Khudzie Mlambo

   Clinton Health Acccess Initiative, Boston, United States

   Contribution

   Data curation, Project administration

   Competing interests

   No competing interests declared

4. Fiona J Walsh

   Clinton Health Acccess Initiative, Boston, United States

   Contribution

   Data curation, Project administration

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2282-1005

5. Emma Mafara

   Clinton Health Acccess Initiative, Boston, United States

   Contribution

   Data curation, Project administration

   Competing interests

   No competing interests declared

6. Charlotte Lejeune

   Clinton Health Acccess Initiative, Boston, United States

   Contribution

   Data curation, Project administration

   Competing interests

   No competing interests declared

7. Cebele Wong

   Clinton Health Acccess Initiative, Boston, United States

   Contribution

   Data curation, Project administration

   Competing interests

   No competing interests declared

8. Anita Hettema

   Clinton Health Acccess Initiative, Boston, United States

   Contribution

   Data curation, Project administration

   Competing interests

   No competing interests declared

9. Osondu Ogbuoji

   Center for Policy Impact in Global Health, Duke Global Health Institute, Duke University, Durham, United States

   Contribution

   Conceptualization, Formal analysis, Writing - review and editing

   Competing interests

   No competing interests declared

10. Sebastian Vollmer

    University of Goettingen, Goettingen, Germany

    Contribution

    Writing - review and editing

    Competing interests

    No competing interests declared

11. Jan-Walter De Neve

    Heidelberg Institute of Global Health, Heidelberg University, Heidelberg, Germany

    Contribution

    Conceptualization, Supervision, Methodology, Writing - review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0003-0090-8249

12. Sikhathele Mazibuko

    Ministry of Health of the Kingdom of Eswatini, Mbabane, Eswatini

    Contribution

    Conceptualization, Data curation, Validation, Investigation, Project administration

    Competing interests

    No competing interests declared

13. Velephi Okello

    Ministry of Health of the Kingdom of Eswatini, Mbabane, Eswatini

    Contribution

    Conceptualization, Data curation, Validation, Investigation, Project administration

    Competing interests

    No competing interests declared

14. Till Bärnighausen

    Heidelberg Institute of Global Health, Heidelberg University, Heidelberg, Germany

    Contribution

    Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Investigation, Methodology

    Competing interests

    No competing interests declared

15. Pascal Geldsetzer

    1. Heidelberg Institute of Global Health, Heidelberg University, Heidelberg, Germany
    2. Division of Primary Care and Population Health, Department of Medicine, Stanford University, Stanford, United States

    Contribution

    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing - original draft, Writing - review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-8878-5505

• 935

  views

• 123

  downloads

• 6

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.