Skip to main content

Table 2 Description of multiple imputation approaches considered to handle missing covariate data

From: Evaluation of multiple imputation approaches for handling missing covariate information in a case-cohort study with a binary outcome

Method*

Accommodation of Weighting in MI

MI Framework

Label

Complete case

No imputation completed. Analysis applied to observations with complete covariate data.

N/A

CCA

Weight only

Imputation models include weights (through the outcome) as a predictor of missingness

FCS

FCS-WO

MVNI

MVNI-WO

Weight interactions

Interaction between outcome (proxy for weight) and exposure/covariates included in imputation model through passive imputation (FCS) or ‘just another variable’ (MVNI), in addition to outcome as a predictor.

FCS

FCS-WX

MVNI

MVNI-WX

Stratum specific imputation

Covariates imputed separately by weight status

FCS

FCS-SS

MVNI

MVNI-SS

Weighted model

Imputation model weighted with inverse probability of selection, outcome included as a predictor.

FCS

FCS-WM

  1. *All methods involve using multiple imputation to address the missing covariates, excluding the complete case analysis, with a weighted analysis model to address the unequal probabilities and missing exposure
  2. FCS Fully Conditional Specification, MVNI Multivariate Normal Imputation, MI Multiple Imputation