Table 3.
Average maximum tumor diameter (cms) of screen-detected and interval cancers by tertiles of mammographic measures
| Mammographic measure | Cancer detection | Least squares means (cms)a(95% CI) by tertiles of mammographic measures | |||
|---|---|---|---|---|---|
| Low | Middle | High | P value b | ||
| Percent density | All | 1.29 (1.16, 1.42) N?=?277 | 1.46 (1.32, 1.60) N?=?269 | 1.45 (1.32, 1.60) N?=?275 | 0.34 |
| Screen detected | 1.25 (1.13, 1.39) N?=?260 | 1.36 (1.23, 1.51) N?=?228 | 1.36 (1.22, 1.51) N?=?213 | 0.16 | |
| Interval cancers | 1.50 (1.06, 2.13) N?=?17 | 2.02 (1.61, 2.54) N?=?41 | 1.87 (1.55, 2.26) N?=?62 | 0.09 | |
| P value c | 0.324 | 0.002 | 0.003 | ||
| Dense area | All | 1.33 (1.20, 1.47) N?=?275 | 1.35 (1.22, 1.48) N?=?273 | 1.50 (1.37, 1.65) N?=?273 | 0.12 |
| Screen detected | 1.27 (1.15, 1.41) N?=?247 | 1.27 (1.15, 1.40) N?=?236 | 1.43 (1.29 1.50) N?=?218 | 0.43 | |
| Interval cancers | 1.66 (1.26, 2.19) N?=?28 | 1.98 (1.56, 2.52) N?=?37 | 1.87 (1.53, 2.27) N?=?55 | 0.06 | |
| P value c | 0.068 | 0.001 | 0.017 | ||
| Non-dense area | All | 1.39 (1.25, 1.54) N?=?276 | 1.45 (1.31, 1.60) N?=?271 | 1.35 (1.22, 1.51) N?=?274 | 0.93 |
| Screen detected | 1.31 (1.17, 1.46) N?=?216 | 1.39 (1.25, 1.55) N?=?227 | 1.29 (1.16, 1.44) N?=?258 | 0.78 | |
| Interval cancers | 1.84 (1.51, 2.24) N?=?60 | 1.81 (1.45, 2.25) N?=?44 | 2.06 (1.43, 2.96) N?=?16 | 0.78 | |
| P value c | 0.002 | 0.031 | 0.013 | ||
aLeast squares means were obtained from linear regression models with an interaction between cancer detection method and tertiles of mammographic measures. Tumor diameter in cms was log transformed for the analysis. Least squares means and their 95% confidence intervals (CIs) are shown transformed back to the original scale. For N?=?15 subjects with recorded tumor size?=?0, the minimum recorded tumor size?=?0.1 cm was used in the analysis. For N?=?17 subjects with screen detected, and for N?=?5 subjects with interval breast cancer the tumor diameter was not available. For tertiles of percent mammographic density, the model was adjusted for age, body mass index, and menopausal status. For tertiles of dense area, the model was, in addition, adjusted for non-dense area (continuous), and for tertiles of non-dense area, the model was adjusted for dense area (continuous). bP values for the test for linear trend across tertiles of mammographic measurements. cP values for comparing least square means of maximum tumor diameter in screen-detected and interval breast cancers within tertiles of mammographic measures. The P values were adjusted for multiple comparisons (the Tukey-Kramer adjustment).