Background:

There are several plausible mechanisms from ART to household welfare, and the direction of net effects is unclear:

1. Patients incur costs when utilizing ART (e.g., for travel, even where ART is free of charge). 1 2. Patients recover health and employment on ART. 2

This study examines the impact of ART on one aspect of household welfare – household food security.

ART may impact household food security in several ways (+/-):

1. Individuals who recover on ART may work more and contribute more to household finances which may be spent on food ( + ) 2. Individuals who recover on ART may regain stamina and contribute more toward household agricultural production ( + ) 3. Individuals on ART may recover appetite faster than stamina putting pressure on limited household resources ( - ) 4. The costs of being on ART, which are high relative to income in this population, may put significant pressure on household resources and

that initially outweigh financial benefits of recovery ( - )

Methods:

Data: Longitudinal health, demographic, and economic data collected from 2004 - 2012 by the Africa Centre for Population Health 3

Sample: 2300 members of the Africa Centre’s Demographic Surveillance Area in rural KwaZulu-Natal, South Africa

Analysis: Regression discontinuity design using earliest CD4 count around the 200 threshold as an instrument for ART initiation 4,5

Model: 𝑌!" = 𝛼!" + 𝛽! ∗ 𝐸𝑥𝑝𝑜𝑠𝑢𝑟𝑒!" + 𝑋!!𝛾 + 𝑇!!𝛿 + 𝜀!"

𝒀𝒊𝒕 is the food security outcome for individual i in survey period t

Exposureit is early ART the instrumented by initial CD4 < 200

Xi represents a vector of individual-specific controls for age at first visit and sex

Tt is a vector of survey year fixed-effects

𝜺𝒊𝒕is an individual specific error term

Note: All standard errors are at the household level & model is run on a bandwidth of 100 CD4 count around CD4 of 200

Outcomes:

1. Probability of an adult in the household missing any food for financial reasons in the last month 2. Probability of an adult in the household missing a meal for financial reasons in the last month 3. Probability of a child in the household missing a meal for financial reasons in the last mont.

Results:

1. ART causes a significant increase in the probability of food insecurity in the year following ART initiation, which diminishes to 0 between 1 and 3 years after ART initiation

2. In the first year after initiation, ART initiation yields a significant increase in: a. The probability of an adult in the household missing any food by 5.5 percentage point (0.055, 95% CI = [0.0190, 0.0904]) b. The probability of an adult in the household missing a meal by 6.5 percentage points (0.065, 95% CI = [0.0156, 0.1147]) c. The probability of a child in the household missing a meal by 4.6 percentage points (0.046, 95% CI = [0.0036, 0.0892])

3. Upper bounds on these causal estimates are an approximately 5 fold increase in household food insecurity due to ART initiation

Conclusions:

1. ART initially places a significant burden on household food security. 2. This negative effect of ART on household food security dissipates over time (between 1 and 3 years (post-initiation) 3. Findings provide evidence for the hypotheses that

a. The financial burden of utilizing ART, which are high relative to income in this community, initially outweigh the longer-term beneficial ART effects on employment and income.

b. Individuals recover appetite at a faster rate than economic benefits, putting pressure on limited household food supply. Policy Recommendation:

Temporary and cost-effective food or financial support programs should be considered to alleviate the short-term loss in food security following ART initiation, especially in the context of the expanding ART rollout and treatment-as-prevention strategies.

References:

1. Chimbindi, N., Bor, J., Newell, M.-L., Tanser, F., Baltussen, R., Hontelez, J., de Vlas, S., Pillay, D., 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 (1999), 70(2), e52–60.

2. Bor, J., Tanser, F., Newell, M.-L., & Bärnighausen, T. (2012). In a study of a population cohort in South Africa, HIV patients on antiretrovirals had nearly full recovery of employment. Health Affairs (Project Hope), 31(7), 1459–1469.

3. Tanser, F., Hosegood, V., Bärnighausen, T., Herbst, K., Nyirenda, M., Muhwava, W., Newell, C., Viljoen, J., Mutevedzi, T., Newell, M.-L. (2008). Cohort Profile: Africa Centre Demographic Information System (ACDIS) and population-based HIV survey. International Journal of Epidemiology, 37(5), 956–962.

4. Bor, J., Moscoe, E., Mutevedzi, P., Newell, M.-L., & Bärnighausen, T. (2014). Regression discontinuity designs in epidemiology: causal inference without randomized trials. Epidemiology (Cambridge, Mass.), 25(5), 729–737.

5. Moscoe, E., Bor, J., & Bärnighausen, T. (2015). Regression discontinuity designs are underutilized in medicine, epidemiology, and public health: a review of current and best practice. Journal of Clinical Epidemiology, 68(2), 122–133.

Prob

abili

ty A

dult

Mis

sed

Food

The Causal Impact of ART Initiation on Household Food Security

Authors: Bryan N. Patenaude1, Natsayi Chimbindi2, Deenan Pillay2,3, Till Bärnighausen1,2

1. Harvard T.H. Chan School of Public Health – Department of Global Health & Population 2. Africa Centre for Population Health 3. University College London – Division of Infection & Immunity

Funding: Data collection with the Africa Centre for Health and Population Studies receives core funding, including for population-based surveillance, from the Wellcome Trust. Till Bärnighausen was supported by a grant from the US National Institute of Child Health and Human Development, R01-HD058482-01. This research was also made possible though the support of the South African Ministry of Health, US Agency for International Development, and the US President’s Emergency Plan for AIDS Relief.

Regression Discontinuity Over 200 CD4 Count Threshold

ART impact on food security

-4

-2

0

2

4

6

8

10

1 2 3 4 5 6 7

-4

-2

0

2

4

6

8

10

1 2 3 4 5 6 7

-4 -2 0 2 4 6 8

10

1 2 3 4 5 6 7

-4 -2 0 2 4 6 8

10

1 2 3 4 5 6 7

-4 -2 0 2 4 6 8

10

1 2 3 4 5 6 7

-4 -2 0 2 4 6 8

10

1 2 3 4 5 6 7

Adult did not eat Adult cut size of meal Child cut size of meal

Δ p

rob

ab

ilit

y

(pp

) Δ

pro

ba

bilit

y

(pp

)

Years on ART

Regression discontinuity at ART eligibility CD4 count cut-offs, controlling for sex and age. N = 1662-2300 for “adult did not eat” and “child cut size of meal”; N = 1662-2297 for “adult cut size of meal”. ITT = intent to treat, CACE = complier average casual effect, pp = percentage points. Patenaude and Bärnighausen CROI 2014

ITT

C

AC

E

Years on ART

Adult Missed Any Food in the Last Month

Child Missed Any Food in the Last Month

ITT

Δ in

Pro

babi

lity

(PP)

C

AC

E

Δ in

Pro

babi

lity

(PP)

Years on ART Years on ART

Years on ART Years on ART

Regression discontinuity at ART eligibility CD4 count cut-offs, controlling for sex and age. N = 1662-2300 for “child missed any food”; N = 1662-2297 for “adult missed any food”. ITT = intent to treat, CACE = complier average casual effect, pp = percentage points.

CD4 Count