Estimation of a Relative Risk Effect Size when Using Continuous Outcomes Data: An Application of Methods in the...

Area of Science:

• Health economics
• Biostatistics
• Psychiatric research

Background:

• Continuous effect size measures like raw mean difference (RMD) and standardized mean difference (SMD) are difficult to integrate into health care decision-analytic models.
• Relative risk (RR) is a more usable effect size for such models, necessitating methods to convert continuous data to RR.

Purpose of the Study:

• To compare the predictive performance of three methods for converting continuous outcomes data from psychiatric rating scales into relative risk (RR) effect sizes.
• To evaluate the utility of these conversion methods in cost-effectiveness models.

Main Methods:

• Three conversion methods (RMD, SMD, Cochrane) were applied to continuous outcomes data from randomized controlled trials (RCTs).
• The RCTs focused on interventions for depression in youth and adults, and eating disorders in young women.
• Predictive performance was assessed using scatterplots, correlation coefficients (r), and linear regression, with an applied cost-effectiveness analysis.

Main Results:

• The RMD and Cochrane methods demonstrated strong predictive performance for RR effect sizes in adult depression and young women's eating disorder RCTs (RMD: r=0.89-0.90; Cochrane: r=0.73-0.96).
• Moderate predictive performance was observed for all three methods in youth depression RCTs (r=0.46-0.50).
• Minimal differences were found when applying the conversion methods within a cost-effectiveness model.

Conclusions:

• The raw mean difference (RMD) and Cochrane conversion methods are valid for predicting relative risk (RR) effect sizes from continuous outcomes data.
• Further validation and refinement are recommended before widespread application of these conversion techniques.