Inspect complete records analysis model

Check complete records analysis is valid under the proposed analysis model and directed acyclic graph (DAG). Validity means that the proposed approach will allow unbiased estimation of the estimand(s) of interest, including regression parameters, associations, and causal effects.

Usage

checkCRA(y, covs = NULL, r_cra, mdag)

Arguments

y: The analysis model outcome variable(s), specified as a string (space delimited) or a list
covs: Optional analysis model covariate(s), specified as a string (space delimited) or a list
r_cra: The complete record indicator, specified as a string
mdag: The DAG, specified as a string using dagitty syntax, or as a dagitty graph object

Value

A message indicating whether complete records analysis is valid under the proposed DAG and analysis model outcome and covariate(s)

Details

The DAG should include all observed and unobserved variables related to the analysis model variables and their missingness, as well as the complete record ("missingness") indicator.

In general, complete records analysis is valid if the analysis model outcome and complete record indicator are unrelated, conditional on the specified covariates. This is determined using the proposed DAG by checking whether the analysis model outcome(s) and complete record indicator are 'd-separated', given the covariates.

References

Hughes R, Heron J, Sterne J, Tilling K. 2019. Accounting for missing data in statistical analyses: multiple imputation is not always the answer. Int J Epidemiol. doi:10.1093/ije/dyz032

Bartlett JW, Harel O, Carpenter JR. 2015. Asymptotically Unbiased Estimation of Exposure Odds Ratios in Complete Records Logistic Regression. Am J Epidemiol. doi:10.1093/aje/kwv114

Examples

# Example DAG for which complete records analysis is not valid
checkCRA(y="bmi7", covs="matage", r_cra="r",
         mdag="matage -> bmi7 mated -> matage mated -> bmi7
               sep_unmeas -> mated sep_unmeas -> r")
#> Based on the proposed directed acyclic graph (DAG), the analysis model
#> outcome(s) and complete record indicator are not independent given
#> analysis model covariate(s). Hence, in general, complete records
#> analysis is not valid.
#> 
#> In special cases, depending on the type of analysis model and estimand
#> of interest, complete records analysis may still be valid. See, for
#> example, Bartlett et al. (2015) (https://doi.org/10.1093/aje/kwv114)
#> for further details.
#> 
#> Consider using a different strategy e.g. multiple imputation, or a
#> different analysis model, noting that a different analysis model may
#> not be aligned with your estimand.
#> 
#> For example, the analysis model outcome(s) and complete record
#> indicator are independent given each of the following sets of
#> variables:
#> 
#> mated
#> 
#> c("matage", "mated")
#> 
#> sep_unmeas
#> 
#> c("matage", "sep_unmeas")
#> 
#> c("mated", "sep_unmeas")
#> 
#> c("matage", "mated", "sep_unmeas")
# For the DAG in the example above, complete records analysis is valid
## if a different set of covariates is used
checkCRA(y="bmi7", covs="matage mated", r_cra="r",
         mdag="matage -> bmi7 mated -> matage mated -> bmi7
               sep_unmeas -> mated sep_unmeas -> r")
#> Based on the proposed directed acyclic graph (DAG), the analysis model
#> outcome(s) and complete record indicator are independent given analysis
#> model covariate(s). Hence, complete records analysis is valid.
# Example where complete records analysis is never valid
checkCRA(y="bmi7", covs="matage mated", r_cra="r",
         mdag="matage -> bmi7 mated -> matage mated -> bmi7
               sep_unmeas -> mated sep_unmeas -> r bmi7 -> r")
#> Based on the proposed directed acyclic graph (DAG), the analysis model
#> outcome(s) and complete record indicator are not independent given
#> analysis model covariate(s). Hence, in general, complete records
#> analysis is not valid.
#> 
#> In special cases, depending on the type of analysis model and estimand
#> of interest, complete records analysis may still be valid. See, for
#> example, Bartlett et al. (2015) (https://doi.org/10.1093/aje/kwv114)
#> for further details.
#> 
#> It is not possible to adjust the analysis model so that the analysis
#> model outcome(s) and complete record indicator are (conditionally)
#> independent.  Consider using a different strategy e.g. multiple
#> imputation.