Table 2 Multivariate logistic regression analysis of HIV infection by trajectory membership, background characteristics and sexual and marital partnership characteristics, Malawi Diffusion and Ideational Change Project, 2001 and 2004
Model 1Model 2
OR (95% CI)OR (95% CI)
Background characteristics
Age (years)0.99 (0.96 to 1.01)0.98 (0.96 to 1.01)
Region
    North (reference)1.001.00
    Central1.24 (0.66 to 2.35)1.07 (0.56 to 2.06)
    South1.87** (1.02 to 3.45)1.72** (0.91 to 3.25)
Trajectory membership
    Sexual debut ⩽15, married 1st sexual partner1.85 (0.98 to 3.49)1.87 (0.98 to 3.58)
    Sexual debut ⩽15, did not marry 1st sexual partner1.43 (0.76 to 2.70)1.22 (0.63 to 2.36)
    Sexual debut >15, married 1st sexual partner (reference)1.001.00
    Sexual debut >15, did not marry 1st sexual partner2.59** (1.30 to 5.16)2.52** (1.25 to 5.10)
Characteristics and behaviours of sexual and marital partners
Total number of lifetime partners
    1 (reference)1.00
    22.24** (1.25 to 4.03)
    ⩾32.56** (1.34 to 4.90)
Was part of least one polygamous union1.54 (0.93 to 2.54)
Had EMSP during any marriage0.94 (0.44 to 2.02)
Knows at least one husband had EMSP1.00 (0.99 to 1.00)
  • CI, confidence interval; EMSP, extramarital sexual partner(s); OR odds ratio.

  • Differences in the likelihood scores between models 1 and 2 with 3 degrees of freedom were statistically significant (difference in 2-log likelihood = 24.37; df = 5; p<0.001). Model 2 therefore provides a better fit.

  • **p<0.05.