Table 2.
Breast Cancer Risk Assessment Tool’s calibration (expected/observed breast cancer cases and 95% confidence intervals) among Nurses’ Health Study (n = 71 293) and the Women’s Health Initiative Extension Study (n = 79 611) participants
| Calibration | NHS* | WHI-ES* | ||||||
|---|---|---|---|---|---|---|---|---|
| Overall (n = 71 293) |
57–74 y† (n = 52 111) |
75+ y (n = 19 182) |
P age comparison | Overall (n = 79 611) |
Age 55–74 y (n = 57 009) |
Age 75+ y (n = 22 602) |
P age comparison | |
| Primary analyses (expected/observed ratios)‡,§ | ||||||||
| BCRAT expected/observed ratios (95% CI) | 1.20 (1.13 to 1.26) |
1.16 (1.09 to 1.23) |
1.31 (1.18 to 1.45) |
.02 | 1.05 (1.00 to 1.10) |
1.03 (0.97 to 1.09) |
1.10 (1.00 to 1.21) |
.21 |
| Hosmer-Lemeshow P | <.0001 | <.0001 | <.0001 | <.0001 | <.0001 | .051 | ||
| Sensitivity analyses | ||||||||
| Limited to women who underwent mammography in past 2 y (NHS = 52 380, WHI = 67 457) | 1.18 (1.11 to 1.25) |
1.17 (1.09 to 1.25) |
1.21 (1.07 to 1.36) |
.58 | 1.04 (0.98 to 1.10) |
1.03 (0.96 to 1.10) |
1.07 (0.97 to 1.18) |
.49 |
| Limited to women without significant illness§ (NHS = 55 567, WHI = 63 439) | 1.19 (1.12 to 1.26) |
1.17 (1.09 to 1.26) |
1.24 (1.09 to 1.40) |
.36 | 1.05 (0.99 to 1.11) |
1.04 (0.97 to 1.11) |
1.07 (0.96 to 1.19) |
.60 |
| Limited to women who underwent mammography in past 2 y and without significant illness, (NHS = 41 887, WHI = 54 255) | 1.17 (1.09 to 1.26) |
1.19 (1.10 to 1.29) |
1.13 (0.98 to 1.30) |
.47 | 1.03 (0.97 to 1.10) |
1.03 (0.96 to 1.11) |
1.04 (0.92 to 1.16) |
.97 |
| Sensitivity analyses (changing SEER incidence)|| | ||||||||
| Using SEER incidence rates from 2006–2010 (NHS = 71 293, WHI = 79 611) |
1.27 (1.20 to 1.34) |
1.25¶ (1.17 to 1.33) |
1.33 (1.20 to 1.48) |
.21 | 1.12 (1.07 to 1.18) |
1.12¶ (1.06 to 1.19) |
1.12 (1.02 to 1.23) |
.97 |
* Nurses’ Health Study included participants alive in 2004. Women’s Health Initiative Extension Study began in 2005. CI = confidence interval; E/O; expected/observed; NHS = Nurses’ Health Study; SEER = Surveillance, Epidemiology, and End Results; WHI-ES = Women’s Health Initiative Extension Study.
† The youngest women in NHS in 2004 were age 57 years.
‡ We compared the expected number of breast cancers based on Breast Cancer Risk Assessment Tool (BCRAT) estimates (calculated using the BCRAT SAS macro) to the observed number in each cohort stratified by age (55–74, 75+ years) (7). To determine the 95% CI of the E/O ratios, we used the Poisson variance of the logarithm of the observed number of cases (19). To test whether E/O ratio estimates differed by age within cohort, we used the normal approximation z-test.
§ Significant illnesses included: history of diabetes, myocardial infarction, congestive heart failure, stroke, emphysema, and peripheral artery disease.
|| For our primary analyses, we used SEER breast cancer incidence rates from 1983–1987 for whites, from 1994–1998 for blacks, from 1990–1996 for Hispanics, and from 1988–2002 for Asians. In sensitivity analyses, we used SEER incidence rates from 2006–2010 for whites, blacks, and Hispanics.
¶ This represents a statistically significant change (P < .05) from the results of the primary analyses. To examine if qualitative differences in E/O ratios resulting from our sensitivity analyses were statistically different than in primary analyses, we used bootstrapping to estimate the standard error of the difference and the z-statistics from which we computed the P values.