The Use of Screening Mammography in Women Aged 70 and Older
The Use of Screening Mammography in Women Aged 70 and Older
Ninety-eight percent of the women in the cohort were white. At the time of entry into the study, 54.6% were aged 70–74, 25.2% were aged 75–79, 13.8% were aged 80–84 and 6.5% were aged 85 and older. Comorbidity scores ranged from 0 to 13.5, but 97% of women had scores of four or lower. Sixty percent of women had none of the health conditions included in the comorbidity index.
Screening mammography use declined steadily with age and was lower with greater comorbidity (Table 1). Logistic regressions with age and comorbidity score as continuous variables indicated that screening declined 9% per year after age 70 (odds ratio (OR) = 0.909, 95% confidence interval (CI) = 0.907–0.911) and was 18% lower with each unit increase in comorbidity score (OR = 0.821, 95% CI = 0.813–0.829). Inclusion of both variables in the same model yielded similar results, and there was no significant interaction, indicating that the age-related decline in screening was independent of comorbidity score and vice versa. The use of diagnostic imaging and breast biopsy in women not subsequently diagnosed with breast cancer generally paralleled the screening rates, except that the benign biopsy rate was not lower with greater comorbidity (Table 1).
Breast cancer incidence (CIS and invasive) in the cohort was 5.8 per 1,000 person-years of follow-up and did not differ significantly with age or comorbidity, despite the differences in screening mammography associated with these variables (Table 1). As a consequence, clinically detected cancers increased 6% per year after age 70 (OR = 1.06, 95% CI = 1.04–1.08) and were 17% greater per unit increase in comorbidity score (OR = 1.17, 95% CI = 1.06–1.28).
Logistic regression results confirmed that having a screening mammogram was associated with a significantly greater likelihood of having diagnostic imaging in the absence of cancer, benign biopsy, or breast cancer diagnosis. For each of these outcomes, ORs (screen vs no screen) were significantly higher than 1 in most age and comorbidity groups (Table 2), but there were some significant interactions. In particular, the ORs for the effect of screening on diagnostic imaging were significantly greater with older age (P < .001 for trend) and greater comorbidity (P = .01). ORs for the effect of screening on benign biopsy were significantly greater with older age (P = .002) but not with greater comorbidity (P = .26). Regression estimates of the yearly probability that a woman without cancer will have diagnostic imaging or a biopsy indicated that these higher the ORs were attributable to larger age-related declines in their usage in unscreened than screened women (Table 3). ORs for the effect of screening on CIS diagnosis were greater with older age, but the trend was not statistically significant (P = .06).
Survival analyses included the 18,356 women with comorbidity data. There were 7,725 deaths during follow-up: 65 of 190 women diagnosed with CIS and 384 of 838 women diagnosed with invasive breast cancer. After adjustment for age, comorbidity, and hospitalization history at the time of entry into the study, CIS was not associated with greater risk of death from any cause (HR = 0.95, 95% CI = 0.75–1.56), but invasive breast cancer during follow-up was significantly associated with greater all-cause mortality (HR = 1.36, 95% CI = 1.22–2.30). The greater risk of all-cause mortality associated with invasive breast cancer was smaller when it was detected through screening (HR = 1.22, 95% CI = 1.07–1.40) than when it was clinically detected (HR = 1.68, 95% CI = 1.43–1.96).
When interaction terms were included in the model, the results indicated that the HRs associated with screen-detected and clinically detected breast cancers were significantly smaller with greater comorbidity (Table 4). Older age at diagnosis was significantly associated with a smaller HR for screen-detected breast cancers but not for clinically detected cancers. These smaller HRs do not correspond to smaller absolute risk of death after diagnosis with invasive breast cancer because baseline mortality is greater with older age and greater comorbidity. For example, in the current study cohort, the estimated 5-year mortality risk for a 70-year-old woman with no serious comorbidity was approximately 3% if she did not have breast cancer and 7% if she had clinically detected invasive breast cancer (a 4-percentage point difference). In contrast, an 85-year-old woman with no serious medical conditions had an approximately 20% chance of dying within 5 years if she did not have breast cancer and a 37% chance of dying if she had clinically detected breast cancer (a 17-percentage point difference). The absolute risk of death for women with screen-detected cancers and no serious comorbidity was 6.3% at age 70 and 23.2% at age 85, so the increase over baseline mortality was nearly the same at both ages. The relationship between detection mode, age at diagnosis, and overall survival is evident in the empirical survival curves shown in Figure 1.
(Enlarge Image)
Figure 1.
Observed survival after diagnosis with screen-detected (——) and clinically detected (– – –) invasive breast cancer and in women of comparable age who were not diagnosed with breast cancer (• • • • • •). Cumulative proportions surviving were obtained from life table analysis and are shown for different ages at diagnosis.
Of the 838 women with invasive breast cancer, 803 had available information for tumor size, 746 for axillary node involvement, and 626 for AJCC stage (Table S2 http://onlinelibrary.wiley.com/store/10.1111/jgs.13184/asset/supinfo/jgs13184-sup-0001-TableS1-S3.doc?v=1&s=02414aff820943a775c2a551efeeb3e6695010c6). Clinically detected cancers were more likely to be larger than 2 cm (OR = 3.53, 95% CI = 2.59–4.83), have lymph node involvement (OR = 2.62, 95% CI = 1.80–3.82), and be Stage II or higher (OR = 3.33, 95% CI = 2.36–4.67) than screen-detected cancers, after adjustment for age. The overall mortality risk in women diagnosed with Stage I breast cancer was not greater than in those who did not develop breast cancer (HR = 0.97, 95% CI = 0.81–1.17) after adjustment for age, comorbidity, and prior hospitalization, but the HR varied with age at diagnosis and comorbidity. The HR was highest for 70-year-old women with no comorbidity (HR=1.71, 95% CI = 1.02–2.13) and decreased by 4% with each year of increase in age at diagnosis and by 12% with each unit increase in comorbidity score. Mortality risk was higher for Stage II (HR = 1.77, 95% CI = 1.49–2.11) and Stage III or IV (HR = 4.96, 95% CI = 2.96–8.30) cancer. There were no significant interactions with age at diagnosis for these higher-stage cancers, but the HRs were significantly lower with greater comorbidity.
Information on treatment and cancer stage was available for 602 of the women with invasive cancer, and it indicated that age at diagnosis highly influenced treatment (Table S3 http://onlinelibrary.wiley.com/store/10.1111/jgs.13184/asset/supinfo/jgs13184-sup-0001-TableS1-S3.doc?v=1&s=02414aff820943a775c2a551efeeb3e6695010c6). Of women aged 70–74, 7.2% of those with Stage I cancer and 7.3% of those with Stage II cancer received only breast-conserving surgery (BCS; excisional biopsy or partial mastectomy without radiation or chemotherapy), whereas of women aged 85 and older, 61.4% of Stage I cancers and 47.2% of Stage II cancers were treated only with BCS. Overall, from age 70 on, the odds of receiving BCS as the only treatment increased 18% per year (OR = 1.18, 95% CI = 1.14–1.23) after adjustment for the effects of tumor stage, comorbidity, and detection mode. Neither detection mode nor comorbidity was related to receiving only BCS, independent of its relationship with age. Of women with Stage I cancers, those who received only BCS had significantly higher overall mortality than women who received other treatment (HR = 2.23, 95% CI = 1.42–3.47), independent of age and comorbidity. For Stage II cancers, overall mortality for women who received only BCS did not differ significantly from that of those receiving additional treatment. The effect of treatment was not examined in the 42 women with Stage III and IV cancers. Because 67.0% of screen-detected cancers were Stage I, those treated only with BCS had poorer overall survival than those receiving other treatments (HR = 1.45, 95% CI = 1.06–1.97).
Results
Participant Characteristics
Ninety-eight percent of the women in the cohort were white. At the time of entry into the study, 54.6% were aged 70–74, 25.2% were aged 75–79, 13.8% were aged 80–84 and 6.5% were aged 85 and older. Comorbidity scores ranged from 0 to 13.5, but 97% of women had scores of four or lower. Sixty percent of women had none of the health conditions included in the comorbidity index.
Screening, Diagnostic Procedures, and Cancer Detection
Screening mammography use declined steadily with age and was lower with greater comorbidity (Table 1). Logistic regressions with age and comorbidity score as continuous variables indicated that screening declined 9% per year after age 70 (odds ratio (OR) = 0.909, 95% confidence interval (CI) = 0.907–0.911) and was 18% lower with each unit increase in comorbidity score (OR = 0.821, 95% CI = 0.813–0.829). Inclusion of both variables in the same model yielded similar results, and there was no significant interaction, indicating that the age-related decline in screening was independent of comorbidity score and vice versa. The use of diagnostic imaging and breast biopsy in women not subsequently diagnosed with breast cancer generally paralleled the screening rates, except that the benign biopsy rate was not lower with greater comorbidity (Table 1).
Breast cancer incidence (CIS and invasive) in the cohort was 5.8 per 1,000 person-years of follow-up and did not differ significantly with age or comorbidity, despite the differences in screening mammography associated with these variables (Table 1). As a consequence, clinically detected cancers increased 6% per year after age 70 (OR = 1.06, 95% CI = 1.04–1.08) and were 17% greater per unit increase in comorbidity score (OR = 1.17, 95% CI = 1.06–1.28).
Logistic regression results confirmed that having a screening mammogram was associated with a significantly greater likelihood of having diagnostic imaging in the absence of cancer, benign biopsy, or breast cancer diagnosis. For each of these outcomes, ORs (screen vs no screen) were significantly higher than 1 in most age and comorbidity groups (Table 2), but there were some significant interactions. In particular, the ORs for the effect of screening on diagnostic imaging were significantly greater with older age (P < .001 for trend) and greater comorbidity (P = .01). ORs for the effect of screening on benign biopsy were significantly greater with older age (P = .002) but not with greater comorbidity (P = .26). Regression estimates of the yearly probability that a woman without cancer will have diagnostic imaging or a biopsy indicated that these higher the ORs were attributable to larger age-related declines in their usage in unscreened than screened women (Table 3). ORs for the effect of screening on CIS diagnosis were greater with older age, but the trend was not statistically significant (P = .06).
Overall Survival
Survival analyses included the 18,356 women with comorbidity data. There were 7,725 deaths during follow-up: 65 of 190 women diagnosed with CIS and 384 of 838 women diagnosed with invasive breast cancer. After adjustment for age, comorbidity, and hospitalization history at the time of entry into the study, CIS was not associated with greater risk of death from any cause (HR = 0.95, 95% CI = 0.75–1.56), but invasive breast cancer during follow-up was significantly associated with greater all-cause mortality (HR = 1.36, 95% CI = 1.22–2.30). The greater risk of all-cause mortality associated with invasive breast cancer was smaller when it was detected through screening (HR = 1.22, 95% CI = 1.07–1.40) than when it was clinically detected (HR = 1.68, 95% CI = 1.43–1.96).
When interaction terms were included in the model, the results indicated that the HRs associated with screen-detected and clinically detected breast cancers were significantly smaller with greater comorbidity (Table 4). Older age at diagnosis was significantly associated with a smaller HR for screen-detected breast cancers but not for clinically detected cancers. These smaller HRs do not correspond to smaller absolute risk of death after diagnosis with invasive breast cancer because baseline mortality is greater with older age and greater comorbidity. For example, in the current study cohort, the estimated 5-year mortality risk for a 70-year-old woman with no serious comorbidity was approximately 3% if she did not have breast cancer and 7% if she had clinically detected invasive breast cancer (a 4-percentage point difference). In contrast, an 85-year-old woman with no serious medical conditions had an approximately 20% chance of dying within 5 years if she did not have breast cancer and a 37% chance of dying if she had clinically detected breast cancer (a 17-percentage point difference). The absolute risk of death for women with screen-detected cancers and no serious comorbidity was 6.3% at age 70 and 23.2% at age 85, so the increase over baseline mortality was nearly the same at both ages. The relationship between detection mode, age at diagnosis, and overall survival is evident in the empirical survival curves shown in Figure 1.
(Enlarge Image)
Figure 1.
Observed survival after diagnosis with screen-detected (——) and clinically detected (– – –) invasive breast cancer and in women of comparable age who were not diagnosed with breast cancer (• • • • • •). Cumulative proportions surviving were obtained from life table analysis and are shown for different ages at diagnosis.
Tumor Characteristics
Of the 838 women with invasive breast cancer, 803 had available information for tumor size, 746 for axillary node involvement, and 626 for AJCC stage (Table S2 http://onlinelibrary.wiley.com/store/10.1111/jgs.13184/asset/supinfo/jgs13184-sup-0001-TableS1-S3.doc?v=1&s=02414aff820943a775c2a551efeeb3e6695010c6). Clinically detected cancers were more likely to be larger than 2 cm (OR = 3.53, 95% CI = 2.59–4.83), have lymph node involvement (OR = 2.62, 95% CI = 1.80–3.82), and be Stage II or higher (OR = 3.33, 95% CI = 2.36–4.67) than screen-detected cancers, after adjustment for age. The overall mortality risk in women diagnosed with Stage I breast cancer was not greater than in those who did not develop breast cancer (HR = 0.97, 95% CI = 0.81–1.17) after adjustment for age, comorbidity, and prior hospitalization, but the HR varied with age at diagnosis and comorbidity. The HR was highest for 70-year-old women with no comorbidity (HR=1.71, 95% CI = 1.02–2.13) and decreased by 4% with each year of increase in age at diagnosis and by 12% with each unit increase in comorbidity score. Mortality risk was higher for Stage II (HR = 1.77, 95% CI = 1.49–2.11) and Stage III or IV (HR = 4.96, 95% CI = 2.96–8.30) cancer. There were no significant interactions with age at diagnosis for these higher-stage cancers, but the HRs were significantly lower with greater comorbidity.
Breast Cancer Treatment
Information on treatment and cancer stage was available for 602 of the women with invasive cancer, and it indicated that age at diagnosis highly influenced treatment (Table S3 http://onlinelibrary.wiley.com/store/10.1111/jgs.13184/asset/supinfo/jgs13184-sup-0001-TableS1-S3.doc?v=1&s=02414aff820943a775c2a551efeeb3e6695010c6). Of women aged 70–74, 7.2% of those with Stage I cancer and 7.3% of those with Stage II cancer received only breast-conserving surgery (BCS; excisional biopsy or partial mastectomy without radiation or chemotherapy), whereas of women aged 85 and older, 61.4% of Stage I cancers and 47.2% of Stage II cancers were treated only with BCS. Overall, from age 70 on, the odds of receiving BCS as the only treatment increased 18% per year (OR = 1.18, 95% CI = 1.14–1.23) after adjustment for the effects of tumor stage, comorbidity, and detection mode. Neither detection mode nor comorbidity was related to receiving only BCS, independent of its relationship with age. Of women with Stage I cancers, those who received only BCS had significantly higher overall mortality than women who received other treatment (HR = 2.23, 95% CI = 1.42–3.47), independent of age and comorbidity. For Stage II cancers, overall mortality for women who received only BCS did not differ significantly from that of those receiving additional treatment. The effect of treatment was not examined in the 42 women with Stage III and IV cancers. Because 67.0% of screen-detected cancers were Stage I, those treated only with BCS had poorer overall survival than those receiving other treatments (HR = 1.45, 95% CI = 1.06–1.97).
Source...