Spectrum 2007 Overview and New Changes UNAIDS Reference Group on Estimates, Model and Projections 2007

Purpose Determine the consequences of the prevalence estimates made with EPP or Workbook –New infections, HIV+ population, AIDS deaths –Need for treatment and effects of treatment –Orphans and vulnerable children

New Features 1.New treatment options Adults: 1 st and 2 nd line ART Children: ART and cotrimoxazole 2.Updated progression periods New infection to need for treatment Need for treatment to AIDS death w/wo treatment 3.Expanded PMTCT options 4.Calibrate to multiple national surveys 5.Uncertainty analysis

Inputs required Country data Demographic data Adult prevalence MTCT program description PMTCT coverage Adult ART coverage Child treatment coverage Epidemic patterns Effect of HIV on fertility Progression from infection to need for treatment and to AIDS death Sex ratio of prevalence Age distribution of infection Mother-to-child transmission rates by regimen and feeding options Effect of child treatment

Creating a new projection

Population Projection

The first year of the projection must be before the start of the AIDS epidemic, usually 1980

Population Projection The last year of the projection can be anything up to 50 years, but it is best to match the prevalence estimate (2012 for EPP, 2007 for the Workbook)

Population Projection

Creates a projection using the UN Population Division’s World Population Prospects

Entering data

Selecting inputs to edit

Adult HIV Prevalence

Documenting sources

Effect of HIV on fertility

15-19: 50% increase 20-49: 30% decrease TFR of HIV- adjusted to leave population TFR unchanged

Age and Sex Ratio of HIV Prevalence

Ratio of prevalence at each to prevalence at : Females

Ratio of prevalence at each to prevalence at : Males

Age and Sex Ratio of HIV Prevalence

Sex ratio of HIV prevalence

Progression from HIV Infection to AIDS Death (Adults) New HIV Infection Need for Treatment First Line ART AIDS Death Second Line ART Need for 2 nd Line

Net survival adjusted to age Source: Zaba and Todd, Alpha network

Time from AIDS or CD4<200 to death in the absence of HAART Time measured from Published studies from resource-poor settings (median, range) Estimates from the Swiss HIV cohort study (median, 95% CI) AIDS CD4 count <200 (median 100) 150< CD4< 250 (median 200) 11 (7 – 19) 11 (7 – 38) 35 (estimated) 17.6 (15.9 – 19) 24.0 (22.0 – 29.0) 77.0 (52.0 – 92.0) Source: Zwahlen and Egger

Time from CD4 < 350 to death, and from CD4 < 350 combined with WHO stage III or IV to death in resource limited settings Time measured from Natural history of untreated patients Bayesian synthesis analysis 200 < CD4 < 350 WHO stage III or IV and CD4 < – 5 years5.4 (1.9 – 15.7 years) 2.7 (0.76 – 9.6 years) Source: Zwahlen and Egger

Incidence (%) of Death After Starting ART Source: Dabis et al, ISPED

Progression from HIV Infection to AIDS Death (Adults) New HIV Infection Need for Treatment First Line ART AIDS Death Second Line ART Normal M 7.5yrs/F 8.5yrs, fast 6.1/6.9 3 years 15%/5% Need for 2 nd Line 15% FY / 5% SY 3 years

Progression from HIV Infection to AIDS Death (Children) New HIV Infection Need for Treatment First Line ARTAIDS Death Age-dependent >5 = 3 years <1, 0.80 >1, 0.9 first year, else 0.95 Marston and Newell, median = 2 years

Adult Male Progression without Treatment

Adult Male Progression with Treatment

The Need for ART will change from 2005 estimate!

Projections of Adult ART Coverage

Orphans % women never married: 38%-94% (71%) % married women in monogamous union: 45%-96% (71%)

Spectrum outputs

Various displays are available

Comparing effects Display up to four projections at one time Open same projection twice and rename to create comparison scenarios

Estimates of Need Adult ARTAdults progressed to need for treatment Simulates WHO guidelines Child ARTChildren progressed to moderate-to- severe disease CotrimoxazoleEarly detection –All HIV+ children No early detection –Children born to HIV+ mothers until 18 months –All HIV+ children over 18 months PMTCTAll births to HIV+ women

Plausibility Bounds

Uncertainty around prevalence curve –EPP: read sampled curve from EPP.spu file –Workbook: generate 1000 logistic curve fits by varying annual estimates according to data quality

Summary Table – Plausibility Bounds

If you use the Workbook you need to generate prevalence curves

Logistic Curve Fits to Annual Estimates with 95% Plausibility Bounds

After generating prevalence curves from Workbook estimates, process the uncertainty analysis

Exercise Create a new demographic projection using EasyProj Read prevalence estimate from EPP or Projections Workbook Select appropriate patterns for Progression and Age/Sex Ratios Add data on coverage of PMTCT, ART and child treatment programs Review key indicators Do uncertainty analysis