Table 2

Results from a stepwise logistic regression* to identify the key socio-demographic characteristics, sexual behaviours and health-related factors associated with high-risk human papillomavirus DNA positivity: women

 % (95% CI) Crude OR (95% CI) Adjusted OR (95% CI) Denominator‡ All 15.9 (14.1 to 17.8) – – 1525, 1851 Age at interview, years p<0.0001 p=0.045 30+ 11.3 (9.5 to 13.3) 1.00 1.00 915, 1146 <30 22.7 (19.4 to 26.4) 2.32 (1.78 to 3.02) 1.37 (1.01 to 1.85) 610, 705 Marital status p<0.0001 p<0.0001 Married 7.2 (5.6 to 9.2) 1.00 1.00 741, 769 Cohabiting 20.7 (15.9 to 26.4) 3.37 (2.21 to 5.15) 2.73 (1.78 to 4.18) 280, 295 Previous married 21.8 (16.3 to 28.6) 3.61 (2.26 to 5.77) 2.68 (1.69 to 4.25) 138, 245 Single, never married 27.5 (23.2 to 32.3) 4.91 (3.49 to 6.90) 3.00 (1.99 to 4.53) 366, 542 Number of new sexual partners,§ last year p<0.0001 p=0.0006 0 11.3 (9.6 to 13.2) 1.00 1.00 1174, 1363 1 29.1 (23.3 to 35.5) 3.22 (2.28 to 4.55) 1.55 (1.08 to 2.22) 200, 293 2+ 42.3 (32.6 to 52.7) 5.77 (3.72 to 8.94) 2.47 (1.53 to 3.99) 98, 138 Concurrency between most recent sexual partners,§ ¶ last year p<0.0001 p=0.032 No 13.9 (12.2 to 15.9) 1.00 1.00 1333, 1599 Yes 31.1 (23.6 to 39.7) 2.78 (1.86 to 4.16) 1.59 (1.04 to 2.44) 101, 144
• * Variables significant at p<0.10 in the bivariate analyses (supplemental table, available online) were considered in the stepwise logistic regression where the significance level for addition to the model is p<0.10. Note that the resulting model also applies when the significance level for addition is p<0.05. When the significance level for addition is p<0.01, there are only two variables in the model: marital status and the ‘number of new opposite/same-sex partners, last year’. No pairwise interactions were statistically significant.

• In order to make the model more statistically efficient, response categories were collapsed where there were no significant differences between categories.

• Weighted, unweighted denominator defined as respondents in Natsal-2 providing a urine sample. Respondents with data missing for 1+ variable(s) were excluded from the multivariate analyses.

• § Opposite and/or same sex sexual partner(s).

• Respondents were asked to report the months and years of first and last sex with their most recent partners (at most three partners). Where these dates overlapped between their most recent partner and their second most recent partner and/or their second most recent partner and their third most recent partner, it is assumed that the partnerships were concurrent.