J
Chron
Dis 1976, Vol. 29. pp. 263-276. Pergamon Press. Printed in Great Britain
FORCES
OF MORTALITY AMONG CANCER PATIENTS
BREAST
RICHARDP. OATES Department
of Preventive Medicine. State University of New York, Upstate Medical Center, 750 East Adams Street. Syracuse, NY 13210 (Receiwd
in rrcrwd
jiirtn
2X Jurw
1975)
INTRODUCTION
IN 1958, Cutler and Ederer Cl] described a life table method for analyzing survival
data. They stressed the usefulness of their method in handling partial survival information. A refinement of this method was reported by Chiang [2] which dealt with the statistical problems encountered when follow-up data are incomplete. The methods presented included (i) an adjustment procedure for use in studies involving survival experience where the cause of death was not specified and (2) an application to studies of mortality from specific causes when competing risks are present. In 1964, Berg [3] proposed “disease-oriented” mortality curves in contrast to the standard patient-oriented survival curves for studies of disease pathology. Berg maintained that the disease-oriented method had not been used as it required information on the causes of death of the individuals under study or at least sufficient knowledge to separate cancer from noncancer deaths. Ederer et al. [4] used cause of death information in a study of mortality among long-term survivors from breast cancer. They found that the excess mortality among breast cancer patients could be explained almost entirely by deaths due to breast cancer. In another long-term follow-up study of breast cancer patients, Brinkley and Haybittle [S] reported that such patients were not more likely than normals to die from causes other than breast cancer. The previous studies of mortality among breast cancer patients have raised other questions regarding the forces of mortality. Are there other causes of death that might be associated with breast cancer? Does age at diagnosis influence the risk of death from causes other than cancer? Are hormonal differences predictive of deaths from malignant neoplasms other than breast cancer’? Age at diagnosis has been an important factor in determining survival. Studies by Berkson et al. [6] and Campos [7] have shown that younger women with axillary metastases have poorer survival rates. and Mersheimer and Heise [S] have 36.1 ( 0 29-I I>
264
RICHARDP. OATES
proposed that breast cancer in younger women may be a different form of the disease than breast cancer in older women. Shapiro et al. [9] reported that case fatality rates among women between 40-49 yr of age were not improved in a breast cancer screening program; whereas, women in the same program in higher decile age groups experienced lower case fatality rates. When the data were combined for all ages, these authors found that approximately one-third of all deaths were due to malignant neoplasms other than breast cancer. Forty-five per cent of their total study sample had causes of death attributable to diseases of the circulatory system. Since there is little information on the forces of mortality among breast cancer patients by age at diagnosis, the study was undertaken. It was hoped that any trends that might appear could be used to help describe and understand the course of mortality of human breast cancer patients.
METHODS
Study population
A total of 2585 female residents of Upstate New York with pathologically confirmed primary diagnoses of breast cancer were recorded by the Upstate Medical Center Cancer Registry from 1956-1970, inclusive. Data were obtained on each patient with regard to age at diagnosis, date of diagnosis, date of last Cancer Registry follow-up or death date, and patient status recorded as dead or alive. Annual follow-up letters are mailed to the patient’s physician on her diagnosis date anniversary. The Upstate Medical Center Cancer Registry attempts to maintain a 99% follow-up on all patients including those that subsequently move out of New York State at any time post-diagnosis. Follow-up data were obtained on each case from the date of diagnosis through April, 1972. The minimum follow-up time post-diagnosis for survivors was one year, since patient status is determined annually. A list of 1218 names of breast cancer patients who had expired in Upstate New York was forwarded to the Bureau of Cancer Control, State of New York, Department of Health, Albany, NY for cause of death information coded in conformity with the International Classification of Diseases (ICD) codes [lo]. Thirty cases died out of state, so that the specific cause of death could only be obtained from the Departments of Health in the states involved. Since these data would be difficult to obtain, the patient records were adjusted to alive status as of the date they moved out of New York State. The amount of time that they survived after leaving New York State was considered censored in the life table analyses. Approximately thirty cases who have moved out of New York State are still alive and being followed by the Upstate Medical Center Cancer Registry. The amounts of time that they have been followed out of state were not removed from the analyses and resulted in a very slight understatement (< 1%) of the relative risks of death described in this paper. Forty-one of the original 1218 deceased individuals were excluded from all analyses as they did not survive at least one month post-diagnosis. Four cases were
Forces of Mortality among Breast Cancer Patients
265
diagnosed at autopsy (two were coded as breast cancer deaths and two were coded as non-cancer deaths). Ten other individuals were found to be terminally ill with conditions other than cancer. They would account for significantly elevated death rates during the first post-diagnostic month. Consequently, it was decided to evaluate the survival of all breast cancer cases beginning 1 month post-diagnosis. This decision resulted in the deletion of 27 cases who were terminally ill with breast cancer. These exclusions resulted in survival analyses based on 1177 dead patients and 1367 live patients for an overall total of 2544 individuals. Statistical
analyses
Life tables. In order to determine possible differences in survival dependent upon age at diagnosis, the study population was stratified into three age groups; ~45. 45-64, 2 65 yr at diagnosis. These age categories were considered in all the analyses which are presented in this paper. Survival curves were obtained for each age group comparing deaths from all causes among the breast cancer population of women versus the life table survival of women of comparable age in the general population of females in Upstate New York (Fig. 1). Age-specific female death rates by cause (Tables 2 and 5) were calculated based on the number of deaths by age and cause as shown in the 81st Annual Statistical Report of the Department of Health, New York State [ 111. Rates were determined by dividing the number of deaths (1960) from a specific cause by the average female population in the specific age group. Since the populations reported in the 1960 United States Census [12] do not extend beyond 85 yr of age, populations in 5 yr age groups between 8.5 and 1OOyr of age were obtained by interpolation. After the age-specific death rates were calculated, life table survival curves were generated for expected deaths from all causes for women of ~45, 45-64, 2 65 yr at diagnosis up to a 15 yr follow-up. Survivals were determined by the statistical procedures inherent in the exposure model described by Sheehe and Feldman [ 133. These methods are illustrated in Appendix item 1 of this paper. The shapes of the distributions of survival times (cases vs age-matched female population) were tested by the chi-square calculation which is inherent in the exposure model [13]. Chi-square contributions were calculated using the number of deaths that occurred during the specific time interval vs the cumulative exposure for that same interval. The number of degrees-of-freedom for the 1’
RICHARD P. OATES
266
tests was based on one less than the number of time intervals in which the amount of exposure exceeded 5. Survival by stage and age at diagnosis
In 1967, the Upstate Medical Center Cancer Registry began staging all carcinomas at the time of diagnosis. Pathology was based on biopsy or mastectomy results. The criteria used for staging were similar to those defined by the EndResults Group of the U.S. National Cancer Institute [14]. For this analysis, all breast cancer cases registered between 1967 and 1972 inclusive were used to obtain the life table survival percentages for all causes of death shown in Table 3. Follow-up on these patients was complete for a full 6-yr period through 1973. The total population for this analysis was 1499 of which 244 were ~45, 736 were 45-64, and 519 were 265 yr of age at diagnosis. Mortality
exposure
One would expect to find differences in survival curves comparing deaths among breast cancer patients vs the general female population. However, if one examined the survival in the two population groups based on causes of death other than malignant neoplasms (OTMN) and deaths from malignant neoplasms other than breast cancer (MNOB) would the two populations have similar survival experiences? For these analyses, mortality exposure measures were calculated for each patient by accumulating the sum of person-years observed within each 5-yr age interval multiplied by the death rates (weighted person-years) for the above causes of death. Relative
risks qf death by age at diagnosis
The relative risks of death (i’s) from OTMN and MNOB were calculated for each of the 3 age at diagnosis groups. The 2 values were obtained by counting the number of deaths observed from OTMN and MNOB and dividing by their respective total exposure measures corresponding to each of the 3 age groups. Chi-square tests employing the maximum In likelihood prodedure were performed to test the hypotheses of equal risks (j’s) within the 3 age groups. Overall chi-square tests were done to compare the general female population versus the breast cancer population with regard to equal risks of death under OTMN and MNOB. The In likelihood procedure is shown under Appendix item 2. Distributions
of’deaths
In order to test the null hypothesis of equal risks of death from OTMN and MNOB, death rates were calculated by 5-yr age intervals (Tables 2 and 5). The categories under OTMN were as follows: (1) allergic, endocrine, metabolic and nutritional diseases; (2) vascular lesions affecting the central nervous system; (3) arteriosclerotic and degenerative heart disease; (4) other heart disease; (5) diseases of arteries and veins and other circulatory system; (6) digestive system; (7) genitourinary; (8) accidents, poisonings and violence and (9) other causes. The categories under MNOB were as follows: (1) digestive organs and peritoneum; (2) respiratory system, (3) genitourinary organs and (4) other neoplasms. The total numbers of deaths observed in each category were compared with expected values (Tables
Forces
of Mortality 1.
TABLE
NUMBER
among
Breast
OF VEATHS AGt
Cancer
OBS~RVHI
RI
261
Patients (‘AI:%
AIVV
AT VIAGNOSIS Age at Diagnosis
Cause
of
< 45
Death
_> 65
45-64
Totals
* Alive at last follow-up.
6 and 7). These were determined by accumulating the total mortality exposure in each category multiplied by the overall risk (3,) of death from OTMN and MNOB. RESULTS
The number of deaths observed by cause and age at diagnosis are shown in Table 1. Seventy-one per cent of all deaths were attributable to breast cancer, while MNOB accounted for 6% and OTMN 23%. For deaths among women diagnosed at ~45 years of age, 93% were attributable to breast cancer. The percentage for those diagnosed at 2 65 yr of age was reduced to 517;. MNOB were given as causes of death for 47% of deaths observed across the three age at diagnosis TABLE
2.
DFATH
RATES*AMONG
FEMALE KCSIDLNTS BY
AC;E
AND
OF UPSTAE
NEW
YORK
(IY60)
CAL’%
RATES? Mallg,,ant
Age
(vi-s.) 100
Neoplasms
:
All But Malignant Neoplasm6
A
B
C
D
E
4.38
0
0
0
0
0
0
0.01 0.01
0.09
0.38 0.29
0
0
0
0
0.06
0. 23
0.52
0
0
0
0.01
0.07
0.44
0.53
0
0
0
0
0.06
0.41
0.72
0.02 0.06
0.01 0.02
0
0.03 0.09
0.08
0.58
1.00
0.10
0.73
1.38
0. 14
0.08
0.11
0. 93
2.26
0.31
0.14
0.18
0.10 0.18
3.72
0.54
0.12
0.31
0.21
2.11
All Causes
0
0.02 0. OS 0.11
0.06
4.28 0.31
1.40
0.62
0. 57
0.06
0. 54
0.33
9.28
0.86
0. 96
0.11
O.Gl
14.57
0.95
1.40
0.19
0.99
0. 48 0.67 0. 86 1.13 1.14 1.34 1.43 0. a0
211.96
2.15
303.72
0
158.02
5.65
23.40
1.09
2.10
0.22
1.30
40.14
1. 34
3.38
0.23
1.54
65.72
1.32
4.25
0. 36
1.52
110.76
1. 55
5.29
0.57
1.73
0.45
2.04
166.35
2.00
6. 16
283.00
1.62
6. 33
0
2.21
313.75
1.43
4.30
0.72
2.36
0
1.43 0
160.38
0
* Per I.OCiI person-years. t R, = D,/P. where R, = death rate for cause A; D, = deaths A [6]; P = population in specific age group. ** A-breast; Baigestive organs and peritoneum: C-respiratory Lgenito-urinary organs; E-other neoplasms.
3.53
6.26 10.37 17.83 32.52 57.13 100.28 154.27
from
CBUSe
system;
RICHARD P. OATES
268
groups. Deaths due to causes OTMN were the highest (42%) among women in the 2 65 yr age group, whereas only 3% were observed in the ~45 yr age group. Survival curves by age at diagnosis for breast cancer cases vs age-matched females of Upstate New York (1960) are shown in Fig. 1. In each of the three age groups, the shapes of the distributions of survival times (cases vs age-matched female population) were highly significantly different. The calculated x2 values were as follows:
< 45
45-64
_> 65
< 45
45-64
265
0.0
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
0. 5
98.37
98.92
96.50
96.14
97.33
97.27
14.58
73.08
53.11
1.0
96.64
97.19
92.03
90.04
90. IO
90.09
60.00
49.79
40.07
1. 5
94.50
95.50
85.90
83.00
al.41
82.45
-
36.00
2.0
91.92
91.90
al.
71.00
76.00
15.00
-
24.03
2.5
86.30
88.44
78.20
12.94
70.94
64.64
-
3. 0
83. a7
86. 26
76.89
65.00
66.00
60.30
-
81
3.5
82.55
83.50
75.00
56.86
57.66
56.46
4.0
82.00
80.50
69.72
55.00
52.57
52.38
-
4.5
79.00
77.50
66.87
47.00
48.00
49.00
-
5.0
77.00
75.18
64.
44.00
43.02
43.
-
5. 5
76.00
72.50
60.29
-
35. a2
41.00
-
6.0
75.45
70.00
55.00
-
31.00
36.00
-
55
16
Total
Deaths
23
65
a6
41
135
91
12
52
Total
Cenaoredt
100
310
210
64
169
92
4
5
Total
Population
123
375
296
105
304
16
57
*In
sifu,
zero
cases
TAlive at last follow-up.
la3
33 7 40
Forces
of Mortality
among
Breast
Cancer
TABLE 4. OBSERVED AND EXPECTED DEATHS AND RELATIVE RISKS OF CAUSES OTHER
NUMBERS OF DEATH FROM
THAN MALIGNANT NEOPLASMS AGE AT DIAGNOSIS
Observed
Chron
Dis 1976, Vol. 29. pp. 263-276. Pergamon Press. Printed in Great Britain
FORCES
OF MORTALITY AMONG CANCER PATIENTS
BREAST
RICHARDP. OATES Department
of Preventive Medicine. State University of New York, Upstate Medical Center, 750 East Adams Street. Syracuse, NY 13210 (Receiwd
in rrcrwd
jiirtn
2X Jurw
1975)
INTRODUCTION
IN 1958, Cutler and Ederer Cl] described a life table method for analyzing survival
data. They stressed the usefulness of their method in handling partial survival information. A refinement of this method was reported by Chiang [2] which dealt with the statistical problems encountered when follow-up data are incomplete. The methods presented included (i) an adjustment procedure for use in studies involving survival experience where the cause of death was not specified and (2) an application to studies of mortality from specific causes when competing risks are present. In 1964, Berg [3] proposed “disease-oriented” mortality curves in contrast to the standard patient-oriented survival curves for studies of disease pathology. Berg maintained that the disease-oriented method had not been used as it required information on the causes of death of the individuals under study or at least sufficient knowledge to separate cancer from noncancer deaths. Ederer et al. [4] used cause of death information in a study of mortality among long-term survivors from breast cancer. They found that the excess mortality among breast cancer patients could be explained almost entirely by deaths due to breast cancer. In another long-term follow-up study of breast cancer patients, Brinkley and Haybittle [S] reported that such patients were not more likely than normals to die from causes other than breast cancer. The previous studies of mortality among breast cancer patients have raised other questions regarding the forces of mortality. Are there other causes of death that might be associated with breast cancer? Does age at diagnosis influence the risk of death from causes other than cancer? Are hormonal differences predictive of deaths from malignant neoplasms other than breast cancer’? Age at diagnosis has been an important factor in determining survival. Studies by Berkson et al. [6] and Campos [7] have shown that younger women with axillary metastases have poorer survival rates. and Mersheimer and Heise [S] have 36.1 ( 0 29-I I>
264
RICHARDP. OATES
proposed that breast cancer in younger women may be a different form of the disease than breast cancer in older women. Shapiro et al. [9] reported that case fatality rates among women between 40-49 yr of age were not improved in a breast cancer screening program; whereas, women in the same program in higher decile age groups experienced lower case fatality rates. When the data were combined for all ages, these authors found that approximately one-third of all deaths were due to malignant neoplasms other than breast cancer. Forty-five per cent of their total study sample had causes of death attributable to diseases of the circulatory system. Since there is little information on the forces of mortality among breast cancer patients by age at diagnosis, the study was undertaken. It was hoped that any trends that might appear could be used to help describe and understand the course of mortality of human breast cancer patients.
METHODS
Study population
A total of 2585 female residents of Upstate New York with pathologically confirmed primary diagnoses of breast cancer were recorded by the Upstate Medical Center Cancer Registry from 1956-1970, inclusive. Data were obtained on each patient with regard to age at diagnosis, date of diagnosis, date of last Cancer Registry follow-up or death date, and patient status recorded as dead or alive. Annual follow-up letters are mailed to the patient’s physician on her diagnosis date anniversary. The Upstate Medical Center Cancer Registry attempts to maintain a 99% follow-up on all patients including those that subsequently move out of New York State at any time post-diagnosis. Follow-up data were obtained on each case from the date of diagnosis through April, 1972. The minimum follow-up time post-diagnosis for survivors was one year, since patient status is determined annually. A list of 1218 names of breast cancer patients who had expired in Upstate New York was forwarded to the Bureau of Cancer Control, State of New York, Department of Health, Albany, NY for cause of death information coded in conformity with the International Classification of Diseases (ICD) codes [lo]. Thirty cases died out of state, so that the specific cause of death could only be obtained from the Departments of Health in the states involved. Since these data would be difficult to obtain, the patient records were adjusted to alive status as of the date they moved out of New York State. The amount of time that they survived after leaving New York State was considered censored in the life table analyses. Approximately thirty cases who have moved out of New York State are still alive and being followed by the Upstate Medical Center Cancer Registry. The amounts of time that they have been followed out of state were not removed from the analyses and resulted in a very slight understatement (< 1%) of the relative risks of death described in this paper. Forty-one of the original 1218 deceased individuals were excluded from all analyses as they did not survive at least one month post-diagnosis. Four cases were
Forces of Mortality among Breast Cancer Patients
265
diagnosed at autopsy (two were coded as breast cancer deaths and two were coded as non-cancer deaths). Ten other individuals were found to be terminally ill with conditions other than cancer. They would account for significantly elevated death rates during the first post-diagnostic month. Consequently, it was decided to evaluate the survival of all breast cancer cases beginning 1 month post-diagnosis. This decision resulted in the deletion of 27 cases who were terminally ill with breast cancer. These exclusions resulted in survival analyses based on 1177 dead patients and 1367 live patients for an overall total of 2544 individuals. Statistical
analyses
Life tables. In order to determine possible differences in survival dependent upon age at diagnosis, the study population was stratified into three age groups; ~45. 45-64, 2 65 yr at diagnosis. These age categories were considered in all the analyses which are presented in this paper. Survival curves were obtained for each age group comparing deaths from all causes among the breast cancer population of women versus the life table survival of women of comparable age in the general population of females in Upstate New York (Fig. 1). Age-specific female death rates by cause (Tables 2 and 5) were calculated based on the number of deaths by age and cause as shown in the 81st Annual Statistical Report of the Department of Health, New York State [ 111. Rates were determined by dividing the number of deaths (1960) from a specific cause by the average female population in the specific age group. Since the populations reported in the 1960 United States Census [12] do not extend beyond 85 yr of age, populations in 5 yr age groups between 8.5 and 1OOyr of age were obtained by interpolation. After the age-specific death rates were calculated, life table survival curves were generated for expected deaths from all causes for women of ~45, 45-64, 2 65 yr at diagnosis up to a 15 yr follow-up. Survivals were determined by the statistical procedures inherent in the exposure model described by Sheehe and Feldman [ 133. These methods are illustrated in Appendix item 1 of this paper. The shapes of the distributions of survival times (cases vs age-matched female population) were tested by the chi-square calculation which is inherent in the exposure model [13]. Chi-square contributions were calculated using the number of deaths that occurred during the specific time interval vs the cumulative exposure for that same interval. The number of degrees-of-freedom for the 1’
RICHARD P. OATES
266
tests was based on one less than the number of time intervals in which the amount of exposure exceeded 5. Survival by stage and age at diagnosis
In 1967, the Upstate Medical Center Cancer Registry began staging all carcinomas at the time of diagnosis. Pathology was based on biopsy or mastectomy results. The criteria used for staging were similar to those defined by the EndResults Group of the U.S. National Cancer Institute [14]. For this analysis, all breast cancer cases registered between 1967 and 1972 inclusive were used to obtain the life table survival percentages for all causes of death shown in Table 3. Follow-up on these patients was complete for a full 6-yr period through 1973. The total population for this analysis was 1499 of which 244 were ~45, 736 were 45-64, and 519 were 265 yr of age at diagnosis. Mortality
exposure
One would expect to find differences in survival curves comparing deaths among breast cancer patients vs the general female population. However, if one examined the survival in the two population groups based on causes of death other than malignant neoplasms (OTMN) and deaths from malignant neoplasms other than breast cancer (MNOB) would the two populations have similar survival experiences? For these analyses, mortality exposure measures were calculated for each patient by accumulating the sum of person-years observed within each 5-yr age interval multiplied by the death rates (weighted person-years) for the above causes of death. Relative
risks qf death by age at diagnosis
The relative risks of death (i’s) from OTMN and MNOB were calculated for each of the 3 age at diagnosis groups. The 2 values were obtained by counting the number of deaths observed from OTMN and MNOB and dividing by their respective total exposure measures corresponding to each of the 3 age groups. Chi-square tests employing the maximum In likelihood prodedure were performed to test the hypotheses of equal risks (j’s) within the 3 age groups. Overall chi-square tests were done to compare the general female population versus the breast cancer population with regard to equal risks of death under OTMN and MNOB. The In likelihood procedure is shown under Appendix item 2. Distributions
of’deaths
In order to test the null hypothesis of equal risks of death from OTMN and MNOB, death rates were calculated by 5-yr age intervals (Tables 2 and 5). The categories under OTMN were as follows: (1) allergic, endocrine, metabolic and nutritional diseases; (2) vascular lesions affecting the central nervous system; (3) arteriosclerotic and degenerative heart disease; (4) other heart disease; (5) diseases of arteries and veins and other circulatory system; (6) digestive system; (7) genitourinary; (8) accidents, poisonings and violence and (9) other causes. The categories under MNOB were as follows: (1) digestive organs and peritoneum; (2) respiratory system, (3) genitourinary organs and (4) other neoplasms. The total numbers of deaths observed in each category were compared with expected values (Tables
Forces
of Mortality 1.
TABLE
NUMBER
among
Breast
OF VEATHS AGt
Cancer
OBS~RVHI
RI
261
Patients (‘AI:%
AIVV
AT VIAGNOSIS Age at Diagnosis
Cause
of
< 45
Death
_> 65
45-64
Totals
* Alive at last follow-up.
6 and 7). These were determined by accumulating the total mortality exposure in each category multiplied by the overall risk (3,) of death from OTMN and MNOB. RESULTS
The number of deaths observed by cause and age at diagnosis are shown in Table 1. Seventy-one per cent of all deaths were attributable to breast cancer, while MNOB accounted for 6% and OTMN 23%. For deaths among women diagnosed at ~45 years of age, 93% were attributable to breast cancer. The percentage for those diagnosed at 2 65 yr of age was reduced to 517;. MNOB were given as causes of death for 47% of deaths observed across the three age at diagnosis TABLE
2.
DFATH
RATES*AMONG
FEMALE KCSIDLNTS BY
AC;E
AND
OF UPSTAE
NEW
YORK
(IY60)
CAL’%
RATES? Mallg,,ant
Age
(vi-s.) 100
Neoplasms
:
All But Malignant Neoplasm6
A
B
C
D
E
4.38
0
0
0
0
0
0
0.01 0.01
0.09
0.38 0.29
0
0
0
0
0.06
0. 23
0.52
0
0
0
0.01
0.07
0.44
0.53
0
0
0
0
0.06
0.41
0.72
0.02 0.06
0.01 0.02
0
0.03 0.09
0.08
0.58
1.00
0.10
0.73
1.38
0. 14
0.08
0.11
0. 93
2.26
0.31
0.14
0.18
0.10 0.18
3.72
0.54
0.12
0.31
0.21
2.11
All Causes
0
0.02 0. OS 0.11
0.06
4.28 0.31
1.40
0.62
0. 57
0.06
0. 54
0.33
9.28
0.86
0. 96
0.11
O.Gl
14.57
0.95
1.40
0.19
0.99
0. 48 0.67 0. 86 1.13 1.14 1.34 1.43 0. a0
211.96
2.15
303.72
0
158.02
5.65
23.40
1.09
2.10
0.22
1.30
40.14
1. 34
3.38
0.23
1.54
65.72
1.32
4.25
0. 36
1.52
110.76
1. 55
5.29
0.57
1.73
0.45
2.04
166.35
2.00
6. 16
283.00
1.62
6. 33
0
2.21
313.75
1.43
4.30
0.72
2.36
0
1.43 0
160.38
0
* Per I.OCiI person-years. t R, = D,/P. where R, = death rate for cause A; D, = deaths A [6]; P = population in specific age group. ** A-breast; Baigestive organs and peritoneum: C-respiratory Lgenito-urinary organs; E-other neoplasms.
3.53
6.26 10.37 17.83 32.52 57.13 100.28 154.27
from
CBUSe
system;
RICHARD P. OATES
268
groups. Deaths due to causes OTMN were the highest (42%) among women in the 2 65 yr age group, whereas only 3% were observed in the ~45 yr age group. Survival curves by age at diagnosis for breast cancer cases vs age-matched females of Upstate New York (1960) are shown in Fig. 1. In each of the three age groups, the shapes of the distributions of survival times (cases vs age-matched female population) were highly significantly different. The calculated x2 values were as follows:
< 45
45-64
_> 65
< 45
45-64
265
0.0
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
0. 5
98.37
98.92
96.50
96.14
97.33
97.27
14.58
73.08
53.11
1.0
96.64
97.19
92.03
90.04
90. IO
90.09
60.00
49.79
40.07
1. 5
94.50
95.50
85.90
83.00
al.41
82.45
-
36.00
2.0
91.92
91.90
al.
71.00
76.00
15.00
-
24.03
2.5
86.30
88.44
78.20
12.94
70.94
64.64
-
3. 0
83. a7
86. 26
76.89
65.00
66.00
60.30
-
81
3.5
82.55
83.50
75.00
56.86
57.66
56.46
4.0
82.00
80.50
69.72
55.00
52.57
52.38
-
4.5
79.00
77.50
66.87
47.00
48.00
49.00
-
5.0
77.00
75.18
64.
44.00
43.02
43.
-
5. 5
76.00
72.50
60.29
-
35. a2
41.00
-
6.0
75.45
70.00
55.00
-
31.00
36.00
-
55
16
Total
Deaths
23
65
a6
41
135
91
12
52
Total
Cenaoredt
100
310
210
64
169
92
4
5
Total
Population
123
375
296
105
304
16
57
*In
sifu,
zero
cases
TAlive at last follow-up.
la3
33 7 40
Forces
of Mortality
among
Breast
Cancer
TABLE 4. OBSERVED AND EXPECTED DEATHS AND RELATIVE RISKS OF CAUSES OTHER
NUMBERS OF DEATH FROM
THAN MALIGNANT NEOPLASMS AGE AT DIAGNOSIS
Observed
