Table 3.
Variable | Incident Detections, No. | Person-Years at Risk, No. | Univariate HR (95% CI) | Incident Detections, No. | Person-Years at Risk, No. | Multivariate HR (95% CI)a |
---|---|---|---|---|---|---|
Age, yb | ||||||
25–34 | 8 | 709 | Reference (1.0) | … | … | … |
35–44 | 8 | 486 | 1.66 (.42–6.57) | … | … | … |
45–54 | 1 | 246 | 0.35 (.05–2.43) | … | … | … |
55–65 | 0 | 190 | 0 (0–2.19)c | … | … | … |
Age at first intercourse, y | ||||||
≤14 | 1 | 125 | Reference (1.0) | … | … | … |
15–16 | 6 | 472 | 1.53 (.23–10.15) | … | … | … |
17 | 1 | 172 | 0.68 (.05–9.25) | … | … | … |
18–19 | 8 | 373 | 2.50 (.32–19.69) | … | … | … |
≥20 | 1 | 490 | 0.24 (.02–3.14) | … | … | … |
Marital status | ||||||
Unmarried or separated | 14 | 1368 | Reference (1.0) | … | … | … |
Married or living with a partner | 2 | 256 | 0.77 (.19–3.16) | … | … | … |
Lifetime male sex partners, no.d | ||||||
1–4 | 0 | 317 | Reference (1.0) | 0 | 317 | Reference (1.0) |
≥5 | 17 | 1277 | ∞ (1.02–∞c) | 16 | 1210 | ∞ (1.02–∞c) |
Currently using hormonal contraceptives | ||||||
No | 11 | 1407 | Reference (1.0) | 11 | 1387 | Reference (1.0) |
Yes | 5 | 140 | 4.66 (1.33–16.38) | 5 | 140 | 4.16 (1.27–13.63) |
Smoking status | ||||||
Never | 10 | 1066 | Reference (1.0) | … | … | … |
Former | 6 | 421 | 1.70 (.44–6.57) | … | … | … |
Current | 1 | 145 | 0.80 (.10–6.52) | … | … | … |
Ever had an abnormal Papanicolaou test resultb | ||||||
No | 9 | 885 | Reference (1.0) | … | … | … |
Yes | 8 | 676 | 1.18 (.33–4.21) | … | … | … |
Analysis was restricted to women reporting no sex with male partners in the 6 months prior to HPV assessment. Women could enter and exit the analysis multiple times if their report of male partners in the prior 6 months varied among follow-up HPV assessments.
Abbreviation: CI, confidence interval.
a Hormonal contraceptive use was adjusted for lifetime number of male sex partners. Lifetime number of male sex partners was unadjusted because we are not aware of exact methods for calculating adjusted CIs.
b A test of scaled Shoenfeld residuals indicated that the proportional hazards assumption was violated. However, no statistically significant variable by time interaction was observed. Therefore, only 1 overall HR is reported.
c Exact unadjusted CI. We are not aware of exact methods for calculating adjusted CIs.
d Lifetime number of male partners was initially categorized into quintiles (1–4, 5–8, 9–14, 15–25, and ≥26); in all analyses, however, similar risks of incident detection among categories of >5 to 8 partners were observed. Therefore, a post hoc decision was made to collapse lifetime number of male partners into a 3-level variable. In this model, lifetime number of partners was further collapsed into a dichotomous variable (1–4 and ≥5) because of 0 outcomes among women with 1–4 partners.