A stepped-wedge randomised trial on the impact of early ART initiation on HIV-patients’ economic outcomes in Eswatini

  1. Janina Isabel Steinert  Is a corresponding author
  2. Shaukat Khan
  3. Khudzie Mlambo
  4. Fiona J Walsh
  5. Emma Mafara
  6. Charlotte Lejeune
  7. Cebele Wong
  8. Anita Hettema
  9. Osondu Ogbuoji
  10. Sebastian Vollmer
  11. Jan-Walter De Neve
  12. Sikhathele Mazibuko
  13. Velephi Okello
  14. Till Bärnighausen
  15. Pascal Geldsetzer
  1. Technical University of Munich, Germany
  2. Clinton Health Acccess Initiative, United States
  3. Center for Policy Impact in Global Health, Duke Global Health Institute, Duke University, United States
  4. University of Goettingen, Germany
  5. Heidelberg Institute of Global Health, Heidelberg University, Germany
  6. Ministry of Health of the Kingdom of Eswatini, Eswatini
  7. Division of Primary Care and Population Health, Department of Medicine, Stanford University, United States
4 figures, 2 tables and 4 additional files

Figures

Participant flow chart (full sample).
Figure 2 with 9 supplements
The causal effect of early ART initiation on economic outcomes.

Notes: Relative Risk presented for negative binomial mixed-effect regression with random intercept by healthcare facility (cluster) and a fixed effect for study period (Hussey and Hughes, 2007). All models control for respondent sex, age, marital status, and highest grade completed and were grand-mean centered. Parametric p-value obtained directly from the regression output; non-parametric p-value obtained from a permutation test with 1000 replications.

Figure 2—figure supplement 1
Histogram: non-resting time use.
Figure 2—figure supplement 2
Histogram: income-generating time use.
Figure 2—figure supplement 3
Histogram: household expenditures.
Figure 2—figure supplement 4
Household assets/living standards.
Figure 2—figure supplement 5
Heterogeneity plots for non-resting time use.
Figure 2—figure supplement 6
Heterogeneity plots for income-generating time.
Figure 2—figure supplement 7
Heterogeneity plots for employment.
Figure 2—figure supplement 8
Heterogeneity plots for household expenditures.
Figure 2—figure supplement 9
Heterogeneity plots for household assets.
Average adjusted predictions of employment rates by period and study arm.

Notes: Percent employed are the average adjusted predictions based on a logistic regression model with a time period fixed effect and a clinic-level random effect, interacting study period with trial arm, and controlling for patients’ age, sex, marital status, level of education, sex (binary), marital status (binary), and their level of education (continuous, specifying the highest grade completed). Period 0 and 7 are not shown because all participants interviewed in period 0 were part of the control phase and all participants interviewed in the last period were exposed to the intervention. The national total labour force participation rate is based on World Bank data and captures the proportion of the population of working age that is economically active during the reference period of 1 year.

Map of the healthcare facilities that participated in the study.

Tables

Table 1
Sample characteristics.
Full study sample (N = 3019)
EAAA
(N = 1868)
SoC
(N = 1151)
Female, n (%)1326 (71.0%)855 (74.3%)
Age, mean (SD)38.3 (11.8)38.3 (11.8)
Education, n (%)
No formal schooling356 (19.1%)212 (18.6%)
Any primary schooling400 (21.4%)294 (25.5%)
Any secondary schooling1112 (59.5%)645 (56.0%)
Married, n (%)1000 (53.5%)651 (56.6%)
Random subsample with data on household expenditure and living standards (N = 1485)
EAAA
(N = 930)
SoC
(N = 555)
Female, n (%)665 (71.5%)417 (75.1%)
Age, mean (SD)38.4 (11.9)38.2 (12.1)
Education, n (%)
No formal schooling175 (18.9%)99 (17.9%)
Any primary schooling192 (20.7%)142 (25.6%)
Any secondary schooling563 (60.5%)314 (56.6%)
Married, n (%)505 (54.3%)316 (56.9%)
Number of household members < 15 years, mean (SD)2.44 (1.11)2.58 (2.00)
Number of household members 15–60 years, mean (SD)2.75 (2.24)3.21 (2.22)
Number of household members > 60 years, mean (SD)0.35 (0.61)0.45 (0.78)
  1. Notes: Abbreviations: EAAA, SD = standard deviation.

Table 2
Stepped-wedge trial design used in this study.
Healthcare facilitySep - Dec
2014
Jan - Apr 2015May - Aug 2015Sep - Dec 2015Jan - Apr 2016May - Aug 2016Sep - Oct 2016Oct 2016 -
Aug 2017
Mshingishingini NazareneCONTINTINTINTINTINTINTINT
NtfonjeniCONTINTINTINTINTINTINTINT
BulandzeniCONTCONTINTINTINTINTINTINT
NdzingeniCONTCONTINTINTINTINTINTINT
MagugaCONTCONTCONTINTINTINTINTINT
MalandzelaCONTCONTCONTINTINTINTINTINT
Pigg's Peak HospitalCONTCONTCONTCONTINTINTINTINT
Peak NazareneCONTCONTCONTCONTINTINTINTINT
HerefordsCONTCONTCONTCONTCONTINTINTINT
Ndvwabangeni NazareneCONTCONTCONTCONTCONTINTINTINT
SigangeniCONTCONTCONTCONTCONTCONTINTINT
SiphocosiniCONTCONTCONTCONTCONTCONTINTNTINT
HoroCONTCONTCONTCONTCONTCONTCONTINT
HhukwiniCONTCONTCONTCONTCONTCONTCONTINT
  1. CONT indicates the control group phase and INT the treatment phase.

Additional files

Source data 1

Datasets, dofiles, and R code for replication purposes.

https://cdn.elifesciences.org/articles/58487/elife-58487-data1-v2.zip
Supplementary file 1

Alternative regression specifications.

(A) The causal effect of EAAA on non-resting time. (B) The causal effect of EAAA on income-generating time. (C) The causal effect of EAAA on employment. (D) The causal effect of EAAA on household expenditures (non-imputed sample). (E) The causal effect of EAAA on household expenditures: Imputed sample. (F) The causal effect of EAAA on asset and living standard index. (G) OLS Specifications. (H) The causal effect of EAAA on a principal component weighted asset and living standard index.

https://cdn.elifesciences.org/articles/58487/elife-58487-supp1-v2.docx
Transparent reporting form
https://cdn.elifesciences.org/articles/58487/elife-58487-transrepform-v2.pdf
Reporting standard 1

CONSORT stepped-wedge trial checklist.

https://cdn.elifesciences.org/articles/58487/elife-58487-repstand1-v2.pdf

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Janina Isabel Steinert
  2. Shaukat Khan
  3. Khudzie Mlambo
  4. Fiona J Walsh
  5. Emma Mafara
  6. Charlotte Lejeune
  7. Cebele Wong
  8. Anita Hettema
  9. Osondu Ogbuoji
  10. Sebastian Vollmer
  11. Jan-Walter De Neve
  12. Sikhathele Mazibuko
  13. Velephi Okello
  14. Till Bärnighausen
  15. Pascal Geldsetzer
(2020)
A stepped-wedge randomised trial on the impact of early ART initiation on HIV-patients’ economic outcomes in Eswatini
eLife 9:e58487.
https://doi.org/10.7554/eLife.58487