Transcript Slides - View the full AIDS 2016 programme

#AIDS2016 | @AIDS_conference
Immediate HIV treatment prevents new infections:
causal evidence on the real-world impact of
immediate vs. deferred ART in rural South Africa
Catherine Oldenburg, Jacob Bor, Frank Tanser, Guy Harling, Tinofa Mutevedzi,
Maryam Shahmanesh, George Seage, Victor De Gruttola, Matthew Mimiaga,
Kenneth Mayer, Deenan Pillay, Till Bärnighausen
#AIDS2016 | @AIDS_conference
Conflicts of Interest
Guy Harling has no conflicts of interest to declare
#AIDS2016 | @AIDS_conference
Benefits from immediate ART
• WHO now recommends immediate ART for
all patients regardless of CD4 count
• Based on evidence from clinical trials
– Prevention
Cohen et al. NEJM 2011, 2016
– Health
Severe et al. 2011; Grinsztejn et al. 2014;
Lundgren et al. 2015; Danel et al. 2015
#AIDS2016 | @AIDS_conference
Little evidence on real world impact
• But potential gap between trial & real-world
– Quality of care
– Generalizability of population
– Behavioral responses to treatment eligibility
#AIDS2016 | @AIDS_conference
Behavioral response to eligibility
Bor et al., CROI 2016
#AIDS2016 | @AIDS_conference
Little evidence on real world impact
• But potential gap between trial & real-world
– Differential quality of care
– Generalizability
– Behavioral responses to treatment eligibility
• In real-world settings, observational data
#AIDS2016 | @AIDS_conference
Incidence in partners
Oldenburg, Bärnighausen, Tanser, Iwuji, De Gruttola, Seage, Mimiaga, Mayer,
Pillay, Harling, Clin Infect Dis, 2016.
#AIDS2016 | @AIDS_conference
Incidence in the household
Vandormael, Newell, Bärnighausen, Tanser. Lancet GH, 2014
#AIDS2016 | @AIDS_conference
Incidence in the community
Tanser, Bärnighausen, Grapsa, Zaidi, Newell. Science, 2013
#AIDS2016 | @AIDS_conference
Little evidence on real world impact
• But potential gap between trial & real-world
– Differential quality of care
– Generalizability
– Behavioral responses to treatment eligibility
• In real-world settings, observational data
– May be affected by unobserved confounders
#AIDS2016 | @AIDS_conference
Study aim
To estimate the causal, real-world effect of
immediate eligibility for ART on:
mortality, immune function,
HIV incidence in the household
We use Regression Discontinuity
to identify causal effects
#AIDS2016 | @AIDS_conference
Study opportunity
• SA National Treatment Guidelines (2004-2011)
– CD4 < 200 or Stage IV
– CD4 ≥ 200 & no Stage IV
#AIDS2016 | @AIDS_conference
Initiate ART
Return in 6 months
Regression discontinuity design
True CD4 count: 200
True CD4 count: 200
True CD4 count: 200
Measured CD4:
188
Measured CD4:
195
Measured CD4:
207
Bor, Mutevedzi, Tanser, Newell, Bärnighausen. Epidemiology, 2014.
Moscoe, Bor, Bärnighausen. Journal of Clinical Epidemiology, 2015.
#AIDS2016 | @AIDS_conference
Study setting
Africa Centre for
Population Health
#AIDS2016 | @AIDS_conference
Study data
• All DSA residents attending public sector clinics
in Hlabisa sub-district
• First CD4 count between Jan 2007 & July 2011
• We see when ART initiated
• We see regular CD4 counts
• Mortality from demographic surveillance
• HIV incidence from population HIV surveillance
#AIDS2016 | @AIDS_conference
CD4 counts at first clinic visit
#AIDS2016 | @AIDS_conference
Continuity in baseline observables
#AIDS2016 | @AIDS_conference
Immediate eligibility raises uptake
#AIDS2016 | @AIDS_conference
Immediate eligibility raises uptake
Risk difference: 0.32
95% CI (0.27, 0.38)
#AIDS2016 | @AIDS_conference
Eligibility reduces mortality
Hazard Ratio = 0.65
95% CI (0.45, 0.94)
Bor, Moscoe, Mutevedzi, Newell, Bärnighausen, Epidemiology, 2014.
#AIDS2016 | @AIDS_conference
Eligibility improves immune health
#AIDS2016 | @AIDS_conference
Persistent impact on CD4 counts
75 cells/μl
#AIDS2016 | @AIDS_conference
ART eligibility reduces
household HIV incidence
Hazard Ratio: 0.55
95%CI (0.35, 0.86)
#AIDS2016 | @AIDS_conference
Potential causal pathways
• Biological: Reduced viral load
• Behavioral: Changes in sexual behavior of
household member on ART
• Relational: Social influence on
other household members
#AIDS2016 | @AIDS_conference
Generalizability
• All effects are local to CD4 count of 200
– May be different at other CD4 count cut-offs
• But similar effect seen for 350 cells/μl cut-off
on ART uptake & retention
Bor et al., CROI 2016a, b
#AIDS2016 | @AIDS_conference
Conclusions
1. Immediate eligibility for ART is associated
with lower mortality, improved immune
function, reduced household HIV incidence
2. Real-world confirmation of trial results &
causal confirmation of observational data
3. Both biological and behavioral pathways
may contribute to the incidence effect
#AIDS2016 | @AIDS_conference
Acknowledgements
• Community and staff at the Africa Centre for
Population Health
• KZN Provincial Department of Health
• Funders:
– Wellcome Trust (097410/Z/11/Z)
– NIH (K01-MH105320, R01-AI112339,
R01-AI124389, R01-HD084233,
R01-HD058482, R25-MH083620, T32-DA013911)
• You can reach the study team at:
[email protected]
[email protected]
#AIDS2016 | @AIDS_conference
[email protected]
[email protected]
Additional material
#AIDS2016 | @AIDS_conference
Robustness to bandwidth
1.6
ITT Hazard Ra o
1.4
Preferred Spec
+/- 150
1.2
1
0.8 ADD FIGURE SHOWING HRRD VS BW
0.6
0.4
0.2
0
175-225
150-250
100-300
50-350
CD4 Count Range (+/- bandwidth)
#AIDS2016 | @AIDS_conference
0-350