Skip to main content

Table 2 Results of mediation analysis testing whether partner communication mediates the effect of type of ANC on partner HIV testing

From: A randomized group antenatal care pilot showed increased partner communication and partner HIV testing during pregnancy in Malawi and Tanzania

 

Non-mediation model

Mediation Models

Model 1.

Effect of Type of ANC on Partner HIV Testing (c)

Model 2.

Effect of Type of ANC on Partner Communication (a)

Model 3.

Effect of Partner Communication (b) and Type of ANC on Partner HIV Testing (c ′ )

Estimate (SE)

OR (CI)

Estimate (SE)

OR (CI)

Estimate (SE)

OR (CI)

Partner Communication, late pregnancy

  

0.45* (0.18) b

1.58 (1.10, 2.26)

Type of ANC (Group vs Individual [ref])

0.69** (0.31)c

1.99 (1.08, 3.66)

1.43*** (0.32)a

4.17 (2.22, 7.83)

0.44 (0.33)c’

1.55 (0.81, 2.95)

Co-variates

 Country (Tanzania vs Malawi [ref])

0.87* (0.32)

2.38 (1.28, 4.43)

1.73*** (0.34)

5.67 (2.93, 10.95)

0.58 (0.34)

1.79 (0.92, 3.47)

 Relationship Status (if married or living with partner vs not in relationship [ref])

1.44* (0.60)

4.24 (1.31, 13.72)

0.87 (0.60)

2.39 (0.74, 7.69)

1.36* (0.61)

3.90 (1.18, 12.83)

  1. p-values: < 0.05 * < 0.01 ** < 0.001 ***​
  2. Note: Indirect Effect: ab = 1.43*0.45 = 0.65; Sobel Test for Indirect Effect: SE = 0.30, z statistics = 2.16, p-value = 0.03