Missing-data strategy

Expected missingness → reasons and assumptions → handling method → sensitivity plan

Flow diagram

Missing-data strategyMissing-data strategyF1.S1Expected missingness and datacollectionF1.S2Intercurrent events versus missingdataF1.S3Missingness assumptionsF1.S4Primary handling methodF1.S5Sensitivity and tipping-pointscenariosF1.S6Report assumptions and implicationsF1.N1Missing-data scenario specificationVariable orendpointMissingnessreasonAssumptionMethodSensitivityparameterSourceEndpointF1.N2Method specificationF1.N3Missingness evidence

Steps

Expected missingness and data collection

Identify variables, times and reasons for missing observations.

Intercurrent events versus missing data

Separate events affecting the estimand from inability to observe required data.

Missingness assumptions

State plausibility of assumptions and information available to assess them.

Primary handling method

Specify likelihood, imputation, weighting or other method with required inputs.

Sensitivity and tipping-point scenarios

Explore departures from the primary assumptions while maintaining the target estimand.

Report assumptions and implications

Document what is observed, assumed and learned from sensitivity analyses.

Tables and supporting information

Missing-data scenario specification

Missing-data scenario specificationVariable orendpointMissingnessreasonAssumptionMethodSensitivityparameterSourceEndpoint

No missingness mechanism or imputation method is assumed by default.

Method specification

Enter method equations only after model and assumptions are specified.

Missingness evidence

Reference anticipated patterns, prior studies and prevention measures.

Sources