The impact of PEPFAR PMTCT funding - PowerPoint Presentation

Slide1

The impact of PEPFAR

PMTCT funding on infant mortality and antenatal care uptake in Kenya: A quasi-experimental evaluation

Dale A. Barnhart,

1

Isaac Tsikhutsu,

2,5

Fredrick Sawe,

2,5

Jane Muli,

2,5

Duncan Kirui,

2,5

William Sugut,

2,5

Nareen

Abboud,

3

 Tiffany Hamm,

4,5

 Peter Coakley,

4,5

Patrick W. Hickey,

5

Vanessa

Wolfman

,

4,5

Elizabeth Lee,

5

Donna Spiegelman,

1

1

Harvard T.H. Chan School of Public Health;

2

Henry M. Jackson Foundation Medical Research International,

3

Office of the U.S. Global AIDS Coordinator and Health Diplomacy; U.S. Department of State,

4

Henry M. Jackson Foundation for the Advancement of Military Medicine,

5

US Military HIV Research Program, Walter Reed Army Institute of Research

BackgroundBetween 2004-2014, PEPFAR invested almost $60 billion worldwide$248 million for PMTCT

in KenyaInvestment coincided with a halving of under-five mortality rate in KenyaThe direct impact of PEPFAR funding for PMTCT has yet to be determined

BackgroundIs PEPFAR funding for PMTCT

associated with improvements in key PMTCT-related health outcomes in Kenya?

Province-level per-capita PEPFAR funding for PMTCT

Neonatal and infant mortality

HIV counseling, testing, and receipt of test results during ANC

Exposure

Outcomes

Methods: Dose-response designQuasi-experimental design based on the principle that if a causal relationship exists, a higher “dose” of the exposure should lead to a larger “response” in the outcome.

A continuous extension of the difference-in-difference design which supports causal inferenceWithin-province comparisons account for time-invariant confounders. Between-province comparisons account for trends over time

The dose-response curve may exhibit non-linear effects.

Methods: All publicly available data sources

1998

2003

PEPFAR funding for PMTCT

2004

COPs

HIV testing at ANC

(N=21,048)

KAIS

KDHS

KDHS

KAIS

KDHS

2007

2008

2012

2014

Neonatal (N=37,616) and infant (N=30,424) mortality

2003

2008

2012

2014

Outcomes

Exposure

Years in which data were collected

Years in which data are available due to retrospective reporting of birth histories

COPs: Country Operational Plans

KDHS: Kenyan Demographic and Health Survey;

KAIS: Kenyan AIDS Indicator Survey

Methods: Modeling the risk ratio

is the per-capita

funding in year

t

for in

province

k

, lagged by

m

years

is a

function

reflecting a potentially non-linear dose-response curve

g

function

reflecting

potentially non-linear time trends

is the fixed effect for province k

is a

vector of characteristics of individual

i

in sampling cluster

j

in

province

k

at time

t

Household

wealth quintile, water and sanitation access, urban/rural status, and mosquito net ownership

Maternal

education, ethnicity, religion, marital status, age at last birth, parity, and exposure to mass

media

Child’s

sex, birth order, and length of preceding birth interval

R

obust

standard errors to account for correlations

among individuals within

sampling

cluster

j

Results: Infant mortalityDose-response relationship between per capita PEPFAR funding for PMTCT and infant mortality. Each panel presents the number of deaths (n) out of the number of infants with complete exposure data (N).

Going

from the 25th to the 75

th percentile

of funding was associated with a12% reduction in infant mortality after a 1-year lag for annual funding, which was sustained after a 2-year lag (95% CI: 1-21%)

31

% reduction in infant mortality

in the unlagged model for

cumulative funding

(95% CI: 11-46%)

Results: HIV testing at ANCDose-response relationship between per capita PEPFAR funding for PMTCT and HIV Testing at ANC, defined as receiving counseling on PMTCT, being tested for HIV, and receiving the results of this HIV test during ANC.

Going from the 25th to the 75th percentile of funding was associated with a

5% increase in HIV testing at ANC after a 3-year lag for annual funding

(95% CI: 1-8%)

5% increase in HIV testing at ANC after a 1-year lag (95% CI: 2-8%) and a 16% increase after a 3-year lag (95% CI: 9-23%) for cumulative funding

Results: No effect on neonatal mortalityDose-response relationship between per capita PEPFAR funding for PMTCT and neonatal mortality. Each panel presents the number of deaths (n) out of the number of neonates with complete exposure data (N).

Discussion: PEPFAR saved children’s lives!Increased PEPFAR funding for PMTCT was associated with reduced infant mortality and

increased HIV testing at ANC in Kenya.Annual per capita funding was not associated with health outcomes in the year of allocation but became beneficial at later lags.

12% reduction in infant mortality after a 1- and 2-year lags5% increase in HIV testing at ANC after a 3-year lag

Cumulative

funding tended to produce stronger associations after shorter lags.31% decrease in infant mortality in the unlagged model5% increase in HIV testing at ANC after a 1-year lag &

16% increase after a 3-year lag

R

esults were

robust to sensitivity analyses

Assumed province-years

with missing funding data

received

minimum, 25

th

percentile, median, 75

th

percentile, and maximum of observed funding

levels

Discussion: Methodologic relevanceQuasi-experimental dose-response designs are a robust and cost-effective option for program evaluationDonors seeking to evaluate programmatic impact can capitalize on this methodology

Collect data on financial expenditures or programmatic activities disaggregated by geographic region and yearLink disaggregated data to pre-existing data sources such as the DHS, AIS, or PHIA surveys

11

Thank you!

Kenya Medical Research Institute/Military HIV Program-Kenya

Isaac

Tsikhutsu

,

MBChB

, MMED

Duncan

Kirui

Fredrick

Sawe

,

MBChB

, MMED

Jane

Mumbi

, BSN

William SugutHarvard T.H. Chan School of Public HealthDonna Spiegelman, ScD Dale A. Barnhart

Office of the U.S. Global AIDS Coordinator

Nareen

Abboud

, PhD,

MPH

Deborah

Birx

, MD

U.S. Military HIV Research

Program,

Walter Reed Army Institute of

Research

Tiffany Hamm,

PhD

Peter Coakley, MD,

DTM&H

LTC Patrick

W. Hickey

,

MD

Vanessa

Wolfman

, MD,

MPH

Elizabeth Lee, DrPH

This

research was funded by National Institute of Health (grant number 5DP1ES025459-03).The views expressed are those of the

authors and do not necessarily represent the official positions of the United States

Government, the U.S. Army,

or the Department of Defense.Contact information: stdls@hsph.harvard.edu and dab000@mail.harvard.edu

Select ReferencesOffice of the U.S Global AIDS Coordinator. Budget information: Latest Funding. https://www.pepfar.gov/funding/budget/index.htm.Office of U.S. Global AIDS Coordinator. PEPFAR Dashboards.

https://data.pepfar.net/global Office of U.S. Global AIDS Coordinator. Country Operational Plans. https://www.pepfar.gov/countries/cop/index.htm Kenya Central Bureau of Statistics, Kenay

Ministry of Health, ORC Macro. Kenya Demographic and Health Survey 2003. Calverton, Maryland, USA: CBS, MOH, and ORC Macro; 2004.Kenya National AIDS and STI Control Programme (NASCOP). Kenya AIDS Indicator Survey 2007. Nairobi: National Bureau of Statistics [Kenya]; 2009

.

Kenya National Bureau of Statistics, Kenya National AIDS Control Council, Kenya National AIDS/STD Control Programme, Health KMoP, Sanitation, Kenya Medical Research Institute. Kenya Demographic and Health Survey 2008-09 Calverton, Maryland, USA: KNBS and ICF Macro; 2010.Kenya National AIDS and STI Control Programme (NASCOP). Kenya AIDS Indicator Survey 2012: Final Report. Nairobi: NASCOP; 2014.Kenya National Bureau of Statistics, Kenya Ministry of Health, Kenya National AIDS Control Council, Kenya Medical Research Institute, Kenya National Council for Population Development. Kenya Demographic and Health Survey 2014: Final Report. Rockville, MD, USA; 2015.

Select ReferencesHabicht J. Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. International Journal of Epidemiology 1999; 28(1): 10-8.

Fitzmaurice G, Laird N, Ware J. Applied Longitudinal Analysis. 2 ed. Hoboken, New Jersey: John Wiley & Sons, Inc; 2011.Wacholder S. Binomial regression in GLIM: estimating risk ratios and risk differences. Am J Epidemiol 1986; 123(1): 174-84.

Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004;

159(7): 702-6.Durrleman

S, Simon R. Flexible regression models with cubic splines. Stat Med 1989; 8(5): 551-61.Hertzmark E, Li R, Hong B, Spiegelman D. The SAS GLMCURV9 Macro. 2014. https://www.hsph.harvard.edu/donna-spiegelman/software/glmcurv9/.