scientific sessions 2015: hiv estimations and projections 2015
TRANSCRIPT
HIV estimations and projections; an overview of the methodology used in 2015
Dr K.A.M. Ariyaratne Consultant Venereologist | Coordinator Strategic Information
National STD/AIDS Control Programme, Sri Lanka
Objective of this Presentation
• An Overview of the Current HIV estimation method
• What data and assumptions used for Sri Lanka during 2015
Estimated number of adults and children newly infected with HIV 2013
Middle East & North Africa25 000
[14 000 – 41 000]
Sub-Saharan Africa1.5 million
[1.3 million – 1.6 million]
Eastern Europe & Central Asia
110 000 [86 000 – 130 000]
Latin America94 000
[71 000 – 170 000]
Caribbean12 000
[9400 – 14 000]
Total: 2.1 million [1.9 million – 2.4 million]
Asia and the Pacific350 000
[250 000 – 510 000]
North America and Western and Central Europe 88 000
[44 000 – 160 000]
In 2013 there were 2.1 million new HIV infections
Why do we need HIV estimations?
• To provide HIV programme managers with estimated numbers relevant to HIV epidemic
• To use in strategic planning for prevention and care services
• For advocacy purposes: numbers infected, numbers dying
• For impact assessment e.g. Expansion of ART programme
• For global reporting/Comparing country Scenarios
Loss of patients in the AIDS treatment cascade
10th of July 2015
Software used for HIV estimations
About the Spectrum Software• Software developed by Avenir Health (Futures Institute)
and East-West Center. Updated routinely
• Guidance from UNAIDS Reference Group on Estimates Modelling and Projections www.epidem.org
• Meetings to consider issues and refine• Engage with research institutes and individuals to answer
specific questions
• Methods published in peer-reviewed journals every other year
Full model for R-Spline and R-Trend is same except for r(t) calculation
8 Source: East West Center and Futures Institute 2013
Need Quality Data and Good Assumptions
garbage in, garbage out
Outline of the Estimation Process
Demographic Data
Program Statistics
Epidemic Patterns
Surveillance and Survey Data
Demographic and Epidemic Calculations
• Mother-to-child transmission• Child model• Adult model
Prevalence / incidence trend
Results• Number HIV+• New Infections
• AIDS deaths• Need for ART
• Need for PMTCT
Step 1. Creating a projection
Step 2. ART elibgibility Criteria
Step 3. HIV Programme Statistics
• Programme Statistics
1. PMTCT data
2. ART data
3. Child treatment data
Cont., HIV Programme Statistics
Step 4. Calculating Incidence
EPP (Estimation and Projection Package) with Adult (15-49) used.
4.1 Incidence: Configuration1. Define the epidemic structure.
1. Concentrated /Low level epidemic
2. Define the sub-population characteristics1. Estimated population size 2. Whether people are likely to move in and out of this
sub-population3. Enter the estimated time (in years) that a person
spends in that sub-population.
HIV sentinel survey and IBBS data used.
4.2 Sub-populations
1. FSW2. MSM3. Drug users4. Client of FSW5. Male Remaining6. Female Remaining
4.3 Entering surveillance data
HIV sentinel survey and IBBS data used.
4.4 Sub-population sizes
4.5 Sub-population - Turn overs
4.6 Sub-population - Turn overs• Following subpopulations were included
1. Drug Users (time in DU 10 years → male remaining pop.)
2. MSM (No turn over)3. Sex worker clients (time in group 7.5 yrs →
male rem. Pop)4. Sex workers (Time in group 5 yrs → female
remain pop)5. Male remaining population (NA)6. Female remaining population (NA)
4.7 Incidence: Curve fitting• Spectrum 2015 offers four fitting methods
• R-Spline• R-Trend• EPP Classic• Workbook
• One-click fitting• Fit all
• Allows multiple projections to be fit
4.8 Models available in EPP 2015
• R-Spline (default)• Applies a set of smooth mathematical functions to give
smoother incidence curves• R-Trend
• Draws on past experience with trends in r observed in actual epidemics to produce a smooth curve
• EPP Classic• A four parameter model that is useful in low data
situations and produces an epidemic that rises and plateaus
Curve generated for MSM
Calibrating the curves
• FSW no calibration done• Clients of sex workers FSW rate down calibrated
by 0.5• MSM up calibrated by a factor of 1.2• DU no calibrations• Male remaining pop- Urban ANC prevalence used
with a down calibration by a factor of 0.5• Female remaining pop -Urban ANC prevalence
used with a down calibration by a factor of 0.5
Shapes of final curves fitted
Getting Results with Spectrum
Reported Vs. Estimated numbers
Reported Numbers - 2014
• Number living with HIV 1737 (Reported – Deaths) (53% of est.)
• New infections in 2014 228
• Deaths in 2014 26
• Adult HIV prevalence 0.03%
Estimated numbers - 2014
• Number living with HIV 3300
• New infections in 2014 <500
• Deaths in 2014 <200
• Adult HIV prevalence <0.1%
KAM Ariyaratne/ National STD/AIDS Control Programme/21.09.2015
Outputs1. HIV pop by risk groups
Outputs: 2. New HIV infections.
Too many DU new infections than actual data
Summary
• Spectrum software used in the recent HIV estimation
• Software is getting improved over the time and subjected
to very frequent software updates
• Methodology is less sensitive to very low prevalent
epidemic
• Better programmatic data, surveillance data and beter
assumptions will produce reasonable estimates
Thank you !