Overview

1
Overview

• Aim offer alternative to CD/UN because they do
  not adequately represent the variation in A-S
  mortality patterns in contemporary populations
  and those with HIV, particularly in the
  developing world
• Result adopt a modified relational system based
  on a universal standard age pattern with two
  age-specific correction factors and two variable
  parameters
• Demonstrate that this largely corrects the
  magnitude-of-deviation-from-the-standard
  dependent distortion inherent in the Brass
  relational logit system

2
Input LTs

• As mentioned, very few from Africa
• Few describe Black populations
• South Africa tables for white population
• 1,802 is a lot
• Would like to see summary tables or plots of the
  quality control measures that were applied

3
General Comments (1)

• Where can be we get ModMatch?
• No attempt to identify families or clusters of
  similar age patterns in the input LTs
• Would like to see a scatter plot of ? v.?? for
  fits to all the input LTs
• Expect some clustering in this plot that would
  correspond to traditional families in other LT
  systems
• Other LT systems have regional or pseudo
  epidemiological frameworks in which to
  interpret families, this system lacks that do
  we simple interpret as a comprehensive set of
  survival curves that have the general form of a
  human survival curve?

4
General Comments (2)

• In the ?, ? two-dimensional space
• Where do CD and UN families live?
• This would be nice in order to relate traditional
  model LT systems to this one and help people
  begin to use them

5
HIV

• As acknowledged, does not have many input LTs
  with HIV
• No single AIDS pattern of mortality lots of
  change as the epidemic matures
• Need at least one more parameter and maybe more
  to account for HIV
• Height of the bump in mortality rates
• Median age of the bump changes as epidemic
  matures
• Width and shape of the bump also changes as
  epidemic matures

6
Male
7
Female