Skip to main content

Table 3 Handling of missing data in primary analysis among the 80 trials who reported missing outcome data

From: Statistical analysis and handling of missing data in cluster randomized trials: a systematic review

Methods

<10 % missing

>10 % missing

Unclear

Total

N = 14

N = 58

N = 8

N = 80

Complete case

10

31

3

44 (55)

Single imputation

    

Worst-case

1

2

0

3 (4)

LOCF

0

2

0

2 (2)

Baseline observation carried forward

0

1

0

1 (1)

Multiple imputation

0

2

0

2 (2)

GEE (unweighted)

3

0

1

4 (5)

Mixed model/hierarchical/multilevel

0

17

1

18 (22)

Othera

0

0

1

1 (1)

Unclear

0

3

2

5 (6)

  1. Abbreviations: LOCF, last observation carried forward; GEE, generalized estimating equation
  2. aOne trial excluded participants who dropped out or had no baseline value; for those who participated at both time points, the LOCF was carried out for a missing primary outcome