Indirect Relative Risk Adjustment (IRRAD)
IRRAD is an Excel spreadsheet that extends the “Axelson adjustment” for binary variables to adjust observed RRs for a (measured or unmeasured) confounding variable when exposure and confounder have multiple categories. IRRAD allows the User to designate a multiplicative or an additive relative risk (RR) model for the joint association of exposure (X) and confounder (C). Additionally, the spreadsheet solves linear equations to define those characteristics that a confounder must have to explain fully or partially an observed association.
- License Agreement
- IndirectRRAdjustment(IRRAD)_NCI_DCEG_ 2018_v1.3.xlsm
- Note: End users will need to enable macros, which are used throughout to create an interactive user interface. All calculations however, are done within the excel worksheets. Updated: December, 2018
The spreadsheet incorporates four approaches to assess confounder effects under either a multiplicative or additive model. These include:
- Binary exposure and binary confounder: a) find the adjusted RR given the observed RR, P(C=1|X=1) and P(C=1|X=0) [or equivalently the observed RR, P(C=1|X=0) and the odds ratio OR(X,C)]; and b) solve for the necessary OR(X,C) given the observed, adjusted and confounder RRs and P(C=1|X=0);
- Categorical exposure and categorical confounder: find the adjusted RRs for each level of exposure given the observed RRs, the confounder distribution at each level of exposure P(C|X) and confounder RRs at the reference level of exposure;
- Categorical exposure and binary confounder: solve for the conditional distribution of the confounder at each level of X, P(C=1|X), given the observed, adjusted and confounder RRs and P(C=1|X=0);
- Categorical exposure and catgorical confounder: solve for the confounder RRs given the observed RRs, the adjusted RRs and P(C|X).
IRRAD includes the data used for the examples in Lubin et al.
Email: Contact DCEG
- Lubin JH, Hauptmann M, Blair A. Indirect adjustment of relative risks of an exposure with multiple categories for an unmeasured confounder. Ann Epidemiol 2018
Note: IRRAD is currently under review and development for 508-compliance.