Skip to main content

Table 4 Mediation analysis of MaxART [4]. (n=1731)

From: Estimating the natural indirect effect and the mediation proportion via the product method

Expression

Scenario

Parameter

Point

S.E.

Delta 95% CI

Bootstrap 95% CI

Steptime adjusted

Approximate

NIE (a)

-0.601

0.091

(-0.779,-0.424)

(-0.782,-0.445)

  

TE (a)

-1.472

0.226

(-1.915,-1.030)

(-1.973,-1.031)

  

MP (a)

0.408

0.069

(0.273,0.544)

(0.305,0.559)

 

Exact

NIE

-0.630

0.093

(-0.813,-0.448)

(-0.816,-0.474)

  

TE

-1.444

0.213

(-1.862,-1.027)

(-1.922,-1.023)

  

MP

0.437

0.073

(0.293,0.580)

(0.327,0.597)

Multivariate adjusted

Approximate

NIE (a)

-0.972

0.121

(-1.208,-0.735)

(-1.287,-0.775)

  

TE (a)

-2.520

0.287

(-3.082,-1.958)

(-3.282,-1.975)

  

MP (a)

0.386

0.050

(0.288,0.483)

(0.292,0.494)

 

Exact

NIE

-0.970

0.120

(-1.205,-0.735)

(-1.282,-0.772)

  

TE

-2.316

0.267

(-2.841,-1.792)

(-3.033,-1.819)

  

MP

0.419

0.054

(0.313,0.525)

(0.322,0.539)

  1. 1Note: All the mediation measures, including NIE, TE, and MP, are defined on a log odds ratio scale for the intervention in change from SoC to EAAA, conditional on the most frequent level of the confounding variables. S.E. denotes the standard error of the point estimates, which is calculated by the multivariate delta method. We implemented the delta method and bootstrap method with 1,000 replications to calculate the 95% confidence interval (95% CI) of each mediation measure