APN 105-Minimum Requirement for Deriving AIDS Extra Mortality Rates

Question 1

Describe the motivation behind APN 105? (4)

Answer 1

• There is sufficient data to illustrate the existence of a wide spread pattern HIV and AIDS epidemic in South Africa
• The potential of the impact is more serious and complex than in Western Europe and North America
• The SAP 104 requires that HIV and AIDS are accounted for in the mortality assumptions when valuing liabilities
• There are many aspects of the HIV & AIDS epidemics which are uncertain particularly the impact of interventions

Question 2

Describe the minimum requirements for determining the HIV preveance and AIDS extra mortality rates? (9)

Answer 2

• Sub-population methodology should be used
o The incidence rates from a population model such as ASSA2008 should be imported into a sub-population mode such as ASSA SELECT model or APN105 model

• The level of risk of HIV infection for a sub-population may be altered by one of the following methods:
o Using risk group incidence rates and then varying the proportion of the sub-population in each risk group (this changes the level and shape of the mortality curve)
o Scaling the incidence rates
o Using aggregated incidence rates from the population model and changing the proportions to approximate the target popuklation

• Multi-stage approach to modelling disease progression is recommended. The disease stage models should include the minimum following states which are used in the APN105 model:
o HIV negative
o HIV positive, asymptomatic
o HIV positive, AIDS related complex (pre-full blown)
o Antiretroviral therapy
o Sick (full blown)

• More complex stages as used by the ASSA SELECT model are possible with appropriate parametization
• Weibull distribution is the preferred method to model transitions as it allows for a time dependent transition rate
• Modelling should produce separate rates for age, sex and calendar year

• The actuary should allow for the effects of selection
o Select periods for HIV selections are fairly long e.g. 15 years
o The dual effect of calendar year and selection needs to be Bourne in mind therefore the rates may need to be dependent on calendar years of selection and policy duration
o Approximations may be acceptable as long as the actuary understands the implications

• Output indicators from the modelling exercise –most commonly aggregate rates should be verified against actual experience
• The parameters of the model should be adjusted in light of credible information

Question 3

List components of epidemiologcal assumptions in the APN 105 model? (4)

Answer 3

Incident rate weighting
Proportion of tagret population assigned to various risk groups

Allow for ART
The time of availability of ARTs made available as well as the proportion in the AIDS sick state who are assumed to initiate ART each year

Passing of HIV test
The user can enter the date in question where a negative HIV result was returned