ARV demand forecast: 2007-2008 Mexican National Institute of Public Health (INSP) Clinton Foundation HIV/AIDS Initiative (CHAI) Geneva, 11 December 2007

1

Background ARV Forecasting Working Group: • • • • • •

UNAIDS WHO National Institute of Public Health, Mexico (INSP) Clinton Foundation HIV/AIDS Initiative (CHAI) USG: SCMS, JSI, MSH, USAID UNICEF

Goal: To develop global ARV demand forecast that is: • Clear and transparent • Easily updated • Sensitive to heterogeneity of epidemiology, protocols, and history of treatment among countries • Realistic 2

Components of the forecasting model

3

Common Assumptions • Scale-up Numbers (linear extrapolation) • Switching from 1st to 2nd line • By cohort, 2% increase in the cumulative switching rate per year for most countries, and 4% for Latin America

4

Patients on treatment (millions and % of total)

5

ARV forecasting: Empirical Approach

Enrollment Data (scale-up curve)

Estimate of new patients added to treatment, assuming linear growth of total on treatment at rate observed in last 12 months Product volumes

Treatment Use Survey (WHO 2006) (snapshot)

Percentage of patients on 1st and 2nd line regimens Sample of 23 countries representing 54% of the LMIC ARV market; extrapolation to 127 LMIC with total treatment data.

6

Daily doses

7

Selected countries (54% of total volume in December 2006) •Burkina Faso •Burundi •Cambodia •Cameroon •Côte d’Ivoire •Ethiopia •Guatemala •Haiti •India •Kenya

•Mozambique •Namibia •Nigeria •Russian Fed. •Rwanda •Swaziland •Tanzania •Thailand •Uganda •Zambia •Zimbabwe

•Lesotho •Malawi 8

ARV forecasting: Normative Approach

Enrollment Data (scale-up curve)

Estimate of new patients added to treatment, assuming linear growth of total on treatment at rate observed in last Patient Population 12 months Product volumes

Protocols (past, current and draft)

Standard 1st and 2nd line regimens (adult/pediatric) and substitutions EPI Data

Model by country for 21 countries representing 83% of the LMIC ARV market 9

ARV forecasting: Normative Approach

Enrollment Data (scale-up curve)

Estimate of new patients added to treatment, assuming linear growth of total on treatment at rate observed in last Patient Population 12 months Product volumes

Protocols (past, current and draft)

Standard 1st and 2nd line regimens (adult/pediatric) and substitutions EPI Data

Model by country for 21 countries representing 83% of the LMIC ARV market 10

ARV forecasting: Normative Approach Patient Population

EPI Data

• • • • • •

Sex Weight Severe peripheral neuropathy Active, diagnosed TB Toxicity-related drug switches Death/program drop-out

• Age • Growth in children • Anemia • Pregnancy • Migration to second line 11

Scope of the forecast Countries included Argentina**

Mozambique*

Botswana*

Namibia**

Brazil**

Nigeria*

Cameroon**

Rwanda*

China*

South Africa**

Côte d’Ivoire**

Tanzania*

Ethiopia*

Thailand**

India*

Uganda**

Kenya*

Zambia**

Malawi*

Zimbabwe**

• These 21 countries represent 83% of global volume • 10 of these countries (33% of global volume) are countries in which CHAI has an office • An additional 10 (48% of global volume) are members of CHAI’s procurement consortium

Mexico * CHAI partner countries **CHAI consortium countries 12

API forecast, 2008 FTC ATV NFV LPV SQV RTV

AZT: normative estimate 1/3 higher than ABC: normative estimate 1/3 higher than empirical empirical estimate: estimate • Implementation regimen exceptions for TDF: normative estimate >5x of empirical women who may become pregnant • 2nd line protocol information estimate

IDV ABC TDF

Empirical Normative

• Move away from D4T-based regimens

ddl

• Recent addition to protocols

EFV AZT NVP 3TC D4T

0

50

100

150

200

250

API volumes, 2008 (metric tons)

300

350 13

Thank you

Clinton Foundation HIV/AIDS Initiative (CHAI) Megan O’Brien [email protected]

Mexican National Institute of Public Health (INSP) Omar Galarraga [email protected]

14

Treatment Attrition • Mortality • Empirical data from Senegal, Brazil, USA • Survival curve derived from cohort studies – AIDS 2006, 20:1181–1189 – The Journal of Infectious Diseases 2006; 194:11–9 – AIDS 2005, 19 (suppl 4):S27–S30

• Lost to follow-up (attrition) • Data from ART-LINC – Lancet 2006; 367: 817–24

15

Planned updates •

Update survival assumptions based on new data

•

Incorporate Dec 2007 treatment data (est. Apr 2008)

•

Include updates need estimates from UNAIDS

•

Market shifts: include expected changes in treatment guidelines and additional product preference scenarios

•

Post forecasts online and allow users to generate subset forecasts • Generic markets • Regional totals • Country subsets

•

Updated survey data from WHO and guidelines information from CHAI

16

Data gaps • Survival on treatment • Survival off treatment • Rates of switch from first to second line • Rates of toxicity-driven switch • Proportion people on treatment that are taking pediatric formulations • In children, distribution among tablets, capsules, and syrup • Distribution of body weight • Prevalence and incidence of pregnancy and TB • Any data contribution or suggestion would be appreciated

17

API forecast (metric tons)

18

PY forecast

19

Data checks Some basic checks 2007: CHAI

AZT or D4T EFV or NVP D4T:AZT ratio NVP: EFV ratio 3TC or FTC: marker for first line Expected first line Difference, actual - expected Difference as % of expected PI total: marker for second line Expected second line Difference, actual - expected Difference as % of expected 3TC or PI: marker for total Expected total Difference, actual - expected Difference as % of expected RTV vs. IND, LPV, SQV, ATV Expected RTV Actual RTV Difference, actual - expected Difference as % of expected

2007: INSP

2008: CHAI

2008: INSP

2,586,000 2,243,469 2.27 3.53 2,289,615 2,235,000 54,615 2.44% 97,768 100,500 -2,732 -2.72% 2,387,383 2,352,500 34,883 1.48%

2,171,689 2,161,187 2.79 3.56 2,192,694 2,235,000 -42,306 -1.89% 126,658 100,500 26,158 26.03% 2,319,352 2,352,500 -33,148 -1.41%

3,279,763 2,873,124 2.25 3.49 2,941,379 2,885,000 56,379 1.95% 147,890 150,000 -2,110 -1.41% 3,089,269 3,035,000 54,269 1.79%

2,854,795 2,839,726 2.78 3.62 2,882,192 2,885,000 -2,808 -0.10% 194,247 150,000 44,247 29.50% 3,076,438 3,035,000 41,438 1.37%

6.74 6.74 0.00 0.00%

9.80 8.70 -1.10 -11.22%

10.32 10.32 0.00 0.00%

13.46 12.00 -1.46 -10.85% 20

PY to persons Can estimate conversion: ((Persons(y)-Persons(y-1)) *0.5)+ Persons(y-1) Ex: Second line • 2006: 81,000 • 2007: 120,000 • 2008: 180,000 PY 2007= ((120,000-81,000)*0.5)+81,000=100,500 PY 2008=((180,000-120,000)*0.5)+120,000=150,000

21

RTV volumes Expected RTV (API)= (IDV/8)+(SQV/10)+(LPV/4)+(ATV/3)

22