Economic differentials in cancer survival: a multivariate analysis.

This study investigates economic differentials in cancer survival using more adequate measures of economic status and controlling for confounding variables more systematically than earlier studies. For 1180 white males, a variant of the Cox regression model is employed to estimate the direct and interaction effects of economic status on survivorship, controlling for age at diagnosis, stage, severity of disease, and initial course of treatment. The results do not show a strong relationship. Estimates of direct or main economic effects rarely reach even borderline statistical significance; they are highly sensitive to model specification and the measurement of the economic variable. An equally weak interaction effect between economic status and stage is detected in several cases, but the parameter estimates are unstable. Such measurement and specification errors have probably exaggerated the importance of economic factors in cancer survival in earlier investigations.