Help needed with stats for dissertation

I posted this earlier in what I now think might have been the wrong area (oops). Aaaand copy and paste:

I'm currently reading many-a-study for my literature review-based dissertation. My stats knowledge from my prior degree is a little rusty and I'm struggling to interpret the analysis in one paper in particular (one of the few findings the author chose not to discuss).

Basically, the study, among many other things, has used self-report questionnaires to look at the influence of demographic, social and psychological factors on the prevalence of self-harm in a cross-section of adolescents. The authors have conducted a logistic regression, univariate and multivariate analysis for each factor.

The factor I am interested in is 'student does not live with biological parent or relative'. The univariate analysis shows an OR of 1.93, p<.01, 95% CI 1.44-2.58 whilst the multivariate analysis shows an OR of 1.27, p is not significant, 95% CI 0.90-1.8.

What does this mean in relation to the correlation between not living with a biological parent or relative and self-harming behaviour? Why is there significance in the univariate but not the multivariate analysis?

Any help would be much appreciated! My poor brain is bamboozled