Table 2

 Logistic regression analysis of inconsistent condom use as a function of statistically significant knowledge, behavioural and psychological variables

VariableRegression coefficient (β)Wald testAdjusted† odds ratio95% CI
†Adjusted for age, nationality, education, occupation, salary, business trips, condom knowledge, having other sex partners, previous STD, condom attitudes, maladaptive behaviours, and all above variables. These variables which were included in the model, but not presented in the table, all had a p value of <0.1 for bivariate analysis and were not statistically significant (ie, p value >0.05) in the regression.
STD knowledge
    10–11 (good)1
    0–9 (poor)1.375.98*3.92(1.31 to 11.68)
No of commercial sex visits
    ⩾50.976.51*2.64(1.25 to 5.55)
Self efficacy −0.145.23*0.87(0.77 to 0.98)
Social norms −0.105.72*0.90(0.83 to 0.98)
Alcohol impaired decision making 0.234.03*1.25(1.01 to 1.56)