– Diagnostic tests • General advice Multivariable Analysis Of Economic Outcomes (I)

• Even if treatment is assigned in a randomized setting use of multivariable analysis may have added benefits: – Improves the power for tests of differences between groups (by explaining variation due to other causes) – Facilitates subgroup analyses for cost-effectiveness (e.g., more/less severe; different countries/centers) – Variations in economic conditions and practice pattern differences by provider, center, or country may have a large influence on costs and the randomization may not account for all differences – Added advantage: Helps explain what is observed (e.g., coefficients for other variables should make sense economically) 2 • If treatment is not randomly assigned, multivariable analysis is necessary to adjust for observable imbalances between treatment groups, but it may NOT be sufficient Multivariable Analysis Of Economic Outcomes (II) • More common techniques – Ordinary least squares regression predicting costs after randomization – Ordinary least squares regression predicting the log transformation of costs after randomizatio