Skip to main content

Table 1 Summary table of proposed general approaches for handling missing participant data for dichotomous outcomes

From: Handling trial participants with missing outcome data when conducting a meta-analysis: a systematic survey of proposed approaches

 

Complete case analysis

Imputations for participants with missing outcome data

Take uncertainty into account

Relation with ITT principle

Assesses risk of bias associated with missing data

Testing of the proposed approach

Based on reasons for missingness

Relative to risk among followed up

Best-case scenario

Worst-case scenario

Other imputation method

Gamble and Hollis [10]

As primary analysis (if missing data non-informative)

As primary analysis (if specified missing data mechanism)

As sensitivity analysis

As sensitivity analysis

Various separate imputations

Handling MPD needed for ITT analysis

Simulation study

Higgins et al. [12]

 

As primary analysis (point of reference)

As primary analysis (preferred)

(Using IMOR)

Applied in 1 meta-analysis of 17 RCTs

Akl et al. [2]

As primary analysis

As primary analysis

Yes (using RILTFU/FU)

As a way to assess risk of bias

Relative to observed incidence in trials included in meta-analysis

Handling MPD differentiated from ITT

Applied in 2 meta-analyses (with 20 and 22 RCTs, respectively)

Mavridis et al. [14]

 

 

 

(Using IMOR)a

Applied in one meta-analysis

  1. RILTFU/FU refers to the event incidence among those lost to follow-up (LTFU) relative to the event incidence among those followed up (FU)
  2. ITT intention to treat, IMOR informative missingness odds ratio