Cost Effectiveness Analysis

• For decision making based on consideration of both consequences or effects and costs
• Measure of effect should relate to the primary objectives of the intervention
• Should comprise a comparison of alternatives
• In these two lies the challenge !

Measures of effect - contrasting views

• Behavioral science - change in behavior and therefore health is the result of a process and this should be acknowledged.
• Health scientists - too often measures of effect have rested in changes in knowledge and attitude that might not translate to health benefits

Aims

• To acknowledge the differing views on measures of effect by representing progressive CE ratios from knowledge change to behavior change to final outcome estimates.
• To complete this for three key themes, HIV/AIDS, Violence Against Women and Hypertension showing estimates of Soul City's cost effectiveness as a prevention intervention for each.

Data Issues

• Two possibilities - Using existing data collected through
  
  - sentinel site surveys
  
  - national surveys.

• Sentinel site data to be used as measured on same subjects - reduces risk of confounding
• Possible problems
  
  - research effect
  
  - generalizability

Key Issues

Possible lack of comparator

• comparator would logically be those not exposed to Soul City
• possible that sample of “comparator group” not large enough and cost outcome description will result
• need to see data but initial feelings are not many cases of this

Whether additional benefit in trying to attribute effect to the different media types

• benefit to Soul City?
• data limitations - no weighting on which media type exposed to most.

Measure of effect

• must be scale measured to result in a sensible “ cost per...”
• coefficients resulting from factor analysis ordinal
• may need to define cut-off for “good” knowledge and “good” behavior and look at % change in sample or additional person showing “good...”