Skip to main content
. 2021 Mar 23;2021:6690275. doi: 10.1155/2021/6690275

Table 2.

Trim-and-fill method for complete remission.

Step 1a

Model Pooled estimate 95% CI Asymptotic
Lower limit Upper limit z value P value Studies (n)
Fixed 0.853 0.595 1.111 6.483 0 23
Random 0.866 0.587 1.144 6.096 0

Test for heterogeneity: Q = 24.331 on 22 degrees of freedom (P=0.330)
Moment-based estimate of between-studies variance = 0.043

Step 2b
Trimming estimator: linear
Meta-analysis type: random effect model

Iteration Estimate Tn # to trim Diff
1 0.866 184 4 276
2 0.796 197 5 26
3 0.79 200 6 6
4 0.784 200 6 0

Step 3c
Filled meta-analysis

Model Pooled 95% CI Asymptotic Studies (n)
Estimate Lower limit Upper limit z value P value
Fixed 0.78 0.529 1.03 6.104 0 29
Random 0.783 0.515 1.051 5.726 0

aMeta-analysis was performed by combined results of all effects with a fixed model and random model. bThe trimming estimator of the random-effect model (through four iterations). cMeta-analysis was performed again after the inclusion of the missing studies.