Forty-two studies (n=1,162,915) were included: 6 RCTs (n=32,194), 15 cohort studies (n=1,097,318) and 21 case-control studies (n=33,403).
There was a linear increase in BC risk by midterm year of case ascertainment (1978 to 2000) based upon data of all study types for MHT (correlation, r=0.593), and to a larger extent for EPT regimens (r=0.664), but not for ET regimens. Annual increases in BC risk for EPT across study types were 0 to 9% and for ET 0 to 3%.
Risk of BC after MHT.
Case-control studies.
There was a statistically significant increased risk of BC associated with MHT in studies with earlier case ascertainment (OR/RR 1.18; 95% CI: 1.09, 1.29; 6 studies); borderline significant heterogeneity was found (P=0.048). There was also a statistically significant increased risk of BC in studies with case ascertainment in 1992 or later (OR/RR 1.34, 95% CI: 1.25, 1.43; 8 studies); significant heterogeneity was found (P=0.008).
Cohort studies and RCTs (n=7) published before 1992.
There was a borderline statistically significantly increased risk of BC associated with MHT in studies with earlier case ascertainment (OR/RR 1.13, 95% CI: 1.00, 1.29; 7 studies); borderline significant heterogeneity was found (P=0.056). There was a statistically significant risk of BC associated with MHT in studies with case ascertainment in 1992 or later (OR/RR 1.70, 95% CI: 1.62, 1.78; 4 cohort studies and 1 RCT); significant heterogeneity was found (P=0.043).
Risk of BC after EPT.
Case-control studies.
There was no statistically significant increased risk of BC associated with EPT in studies with earlier case ascertainment (OR/RR 0.99, 95% CI: 0.84, 1.17; 6 studies); no significant heterogeneity was found (P=0.48). However, there was a statistically significantly increased risk of BC in studies with case ascertainment in 1992 or later (OR/RR 1.48, 95% CI: 1.33, 1.65; 7 studies); no significant heterogeneity was found (P=0.17).
Cohort studies and RCTs.
There was a statistically significantly increased risk of BC associated with EPT in studies with earlier case ascertainment (OR/RR 1.33, 95% CI: 1.14, 1.54; 3 studies); no significant heterogeneity was found (P=0.56). There was also a statistically significantly increased risk of BC in studies with case ascertainment in 1992 or later (OR/RR 1.95, 95% CI: 1.87, 2.04; 3 studies); significant heterogeneity was found (P=0.001).
The analyses reported above were repeated for unopposed oestrogen therapy.