TABLE 4.
Hazard ratios for invasive breast cancer for E+ P* use, in relation to prior hormone therapy and years from E+P therapy initiation, for women who began hormone therapy at menopause (gap time of zero) in the US Women’s Health Initiative clinical trial and observational study (enrollment, 1993–1998)
| No. of years from E+P initiation† | Clinical trial‡ |
Observational study‡,§ |
Combined studies‡,¶ |
|||
|---|---|---|---|---|---|---|
| HR* | 95% CI* | HR | 95% CI | HR | 95% CI | |
| No prior hormone therapy# | ||||||
| <2 | 0.79 | 0.40, 1.57 | 1.69 | 0.60, 4.77 | 0.98 | 0.56, 1.72 |
| 2–5 | 1.97 | 1.11, 3.50 | 1.85 | 1.13, 3.02 | 2.01 | 1.41, 2.86 |
| >5 | 1.99 | 0.92, 4.29 | 2.94 | 2.33, 3.69 | 2.85 | 2.29, 3.54 |
| Prior hormone therapy# | ||||||
| <2 | 1.11 | 0.54, 2.25 | 2.63 | 0.61, 11.29 | 1.28 | 0.66, 2.51 |
| 2–5 | 3.42 | 1.68, 6.95 | 1.64 | 0.63, 4.25 | 2.56 | 1.54, 4.24 |
| >5 | 2.42 | 0.64, 9.17 | 3.32 | 1.79, 6.13 | 3.30 | 1.90, 5.73 |
| Factor for 5-year increase in gap time | 0.84 | 0.69, 1.03 | 0.79 | 0.66, 0.96 | 0.81 | 0.71, 0.91 |
E+P, estrogen plus progestin; HR, hazard ratio; CI, confidence interval.
Time from E+P initiation was defined as time from enrollment for women assigned to E+P in the clinical trial and as the sum of this time plus duration of the ongoing E+P episode at the time of enrollment in the observational study. The number of invasive breast cancer cases among E+P users in the <2, 2–5, and >5 years from E+P initiation were, respectively, 6, 41, and 15 for the clinical trial no prior hormone therapy group; 14, 12, and 5 for the clinical trial prior hormone therapy group; 4, 22, and 156 for the observational study no prior therapy group; and 2, 6, and 25 for the observational study prior hormone therapy group.
HRs and 95% CIs were derived from Cox models that stratified baseline rates on age (5-year categories) and prior hormone therapy status and cohort (clinical trial or observational study). Women for whom age at menopause or age at first use of hormone therapy was missing were omitted, leaving data on 15,139 (91.2%) of the clinical trial women available for analysis.
HR estimates in the observational study controlled for confounding factors separately in the prior hormone therapy and no prior hormone therapy groups and included age (linear), body mass index (<25, 25–29, 30–34, >34 kg/m2, and linear), education (high school or less, beyond high school, college degree), smoking (never, past, current), alcohol intake (never, past, <1/ week, 1–7/week, >7/week), general health (fair/poor, good/very good/excellent), physical activity in metabolic equivalent units/week (0–3.75, 3.76–8.75, 8.76–17.5, >17.5), family history of breast cancer (yes, no), 5-year Gail et al. (17) model breast cancer risk % (<1.25, 1.25–1.74, >1.74, and linear), and bilateral oophorectomy (yes, no). For women with prior hormone therapy, confounding factors also included prior E+P use in years (none, <5, 5–10, >10) and prior estrogen-alone use in years (none, <5, 5–10, >10). Women for whom confounding-factor data, age at menopause, or age at first use of hormone therapy was missing were omitted, leaving data on 27,954 women (87.1% of the observational study subcohort) for analysis.
HR estimates from combined study analyses used the same statistical model as those used in separate clinical trial and observational study analyses but restricted the hazard ratios to be common in the two cohorts up to a multiplicative factor (to control for residual confounding in the observational study). That factor was estimated as 1.03 (95% CI: 0.69, 1.53), indicating excellent overall agreement between hazard ratios in the two cohorts.
Prior (postmenopausal) hormone therapy was defined relative to the baseline E+P episode for E+P users in the observational study and relative to Women’s Health Initiative enrollment otherwise.