American Journal of Epidemiology Copyright © 1992 by The Johns Hopkins University School of Hygiene and Public Health Al rights reserved
Vo) 136, No. 4 Printed in U.S.A.
A BRIEF ORIGINAL CONTRIBUTION
Hakan Olsson1 and Gunnar Hagglund2
At the Department of Orthopedics of University Hospital, Lund, Sweden, cancer incidence and overall and cause-specific mortality were studied in a prospective cohort of 677 women who had experienced a fracture of the distal forearm in 1974 or 1975. The women were followed through population-based population, cancer, and death registries in southern Sweden. There was a significant reduction in overall cancer incidence (66 cases observed vs. 90.21 expected), breast cancer incidence (11 cases observed vs. 20.31 cases expected), and incidence of tumors of the female genital tract (5 cases observed vs. 11.84 expected). Overall mortality was also reduced (146 cases observed vs. 191.69 cases expected), and rates of death from circulatory disease (79 cases observed vs. 111.49 expected) and malignant tumors (30 cases observed vs. 40.41 expected) were both lower than expected. Am J Epidemiol 1992;136:422-7. breast neoplasms; cohort studies; estrogens; fractures; morbidity; mortality; neoplasms; osteoporosis
consequences of estrogen deficiency, such as peri-/postmenopausal symptoms, fragility fractures, and accelerated cardiovascular disease (5-7). The overall benefit would include increased life expectancy. It has been argued that the possible adverse effects of estrogen supplementation—breast and endometrial carcinoma—would not change the overall positive effect of such supplementation (8). While the association between endometrial cancer and estrogen supplementation is established, the question of whether estrogen replacement therapy increases the risk of breast cancer is still controversial. A recent meta-analysis indicates, however, that the association between long-term estrogen supplementation and breast cancer is real (9). We investigated cancer morbidity and total and cause-specific mortality in a cohort of women with distal forearm fractures in order to gain insight into the total picture of morbidity and mortality in women suffering from osteoporosis. Since forearm fractures generally strike women somewhat earlier
Exposure to endogenous and exogenous estrogens has been implicated as a cause of breast cancer and endometriaJ carcinoma (1-4). It is therefore reasonable to hypothesize that medical conditions leading to or indicating estrogen deficiency would at least partially protect against these types of tumors. Such conditions would include early menopause, body leanness (postmenopausal), and signs of accelerated bone loss (osteoporosis). It has been suggested that estrogen substitution peri- and postmenopausally could prevent some of the medical
Received for publication September 3, 1991, and in final form April 6, 1992. Abbreviation: SMR, standardized morbidity/mortality ratio. 1 Department of Oncology, University Hospital, Lund S-221 85, Sweden. 1 Department of Orthopedics, University Hospital, Lund S-221 85, Sweden. Reprint requests to Dr. Hakan CHsson, Department of Oncology, University Hospital, Lund S-221 85, Sweden. This study was supported by grants from the Medical Faculty of Lund University. 422
Downloaded from https://academic.oup.com/aje/article-abstract/136/4/422/164777 by Columbia University user on 17 January 2019
Reduced Cancer Morbidity and Mortality in a Prospective Cohort of Women with Distal Forearm Fractures
Cancer in Women with Forearm Fractures
MATERIALS AND METHODS
Through computerized diagnostic data bases, we identified all women (n = 677) with a first distal forearm fracture seen in the Orthopedic Department of University Hospital, Lund, Sweden, in 1974-1975. The hospital serves the Health District of Lund and has population-based recruitment. The studied population represents all patients who sustained a distal forearm fracture during the given time period in this geographic area. The data base containing the orthopedic diagnoses in 1974-1975 was fully computerized. The median age of the women was 62 years (range, 23-88 years). At the time of the fracture, 24 women were younger than age 30 years, 108 were aged 30-50 years, 378 were aged 50-70 years, and 167 were over age 70 years. By means of the unique Swedish 10-digit personal identification number, we followed the vital status and cancer incidence of these patients from 1974 onward in the population-based Census Registry, Cause of Death Registry, and Swedish Cancer Registry (the South Swedish Regional Tumour Registry and the National Swedish Tumour Registry). Each individual could have had more than one tumor registered. Observed versus expected numbers of cases were then calculated for various tumor types/causes of death usingreferencedata from the southern health care region. The expected morbidity and mortality figures were derived from the same population of women from which the fracture cases were derived. Expected cancer incidence was
calculated using data from the South Swedish Regional Tumour Registry. In the calculations, we used the person-years method, classifying the individuals into 5-year age groups and using a single calendar year as the unit cell size. Vital status was determined up to January 1, 1990. The vital status of the women is presented in table 1 as of January 1, 1987. Five subjects were lost to follow-up because of insufficient information on identity, and three patients were lost because of emigration. Cause-specific standardized morbidity/ mortality ratios and their 95 percent confidence intervals were calculated. The/? values were calculated by use of the Poisson distribution, or the chi-square distribution if the expected values were greater than 10. The term "significant" refers to p < 0.05. All tests were two-tailed. Data on expected cancer morbidity in the health care region were available and updated as of January 1, 1990. Data on expected cancer mortality were available and updated to January 1, 1987. For assessment of cancer morbidity, diseases were grouped according to the International Classification of Diseases, Seventh Revision, and for the mortality analysis, diseases were grouped according to the International Classification of Diseases, Ninth Revision. RESULTS
Table 2 shows cancer morbidity as of January 1, 1990, for various tumor types. During the follow-up period, 66 women developed malignant tumors, compared with an expected incidence of 90.21 (standardized morbidity ratio (SMR) = 0.73, p < 0.01, in comparison with the expected incidence in TABLE 1. Vital status of a cohort of women with distal forearm fractures, South Sweden, as of January 1, 1987 Vital status
No.
Living Deceased Emigrated from Sweden Unknown (insufficient identity)
523 146 3 65
77.3 21.6 0.4 0.7
Downloaded from https://academic.oup.com/aje/article-abstract/136/4/422/164777 by Columbia University user on 17 January 2019
than hip fractures and (contrary to hip fractures) have not previously been suggested to increase mortality (10), we chose to study this group of women to get a clearer picture of the overall health consequences of osteoporosis. From a 1974 study (conducted by absorptiometry) in our health care region of bone mineral content in the forearms of women with distal forearm fractures (11), it was concluded that this group of women had reduced bone mineral content compared with healthy referents.
423
424
Olsson and Hagglund
TABLE 2. Cancer morbidity in a prospective cohort of women who had distal forearm fractures in 1974 or 1975* ICD-7t code(s)
Type of cancer
ot»t
Expt
SMRt
95% a t
P
All malignant tumors
66
90.21
0.73
0.57-0.94
0.01
170 171-176 171 172 175 176 140-148 150 151 152-153 154 155 157 160-164 180-181 190-191 193 194-195 196-197 200-203 204-209
Breast Female genital organs Cervix uteri Uterine body Ovary Vulva/vagina Mouth/pharynx Esophagus Stomach Small and large bowel Rectal carcinoma Liver and biliary tract Pancreas Respiratory tract Urinary organs Skin tumors Brain Endocrine glands Bone/soft tissue sarcoma Lymphoma Leukemia, potycythemia
11 5 1 1 1 0 1 0 3 8 5 3 1 7 4 8 3 2 0 2 1
20.31 11.84 1.75 4.03 4.78 0.84 1.20 0.54 4.07 9.81 4.42 3.67 3.07 3.83 5.19 5.96 2.87 3.00 0.64 3.85 2.24
0.54 0.42 0.57 0.74 0.21 0.00 0.83 0.00 0.74 0.82 1.13 0.82 0.33 1.83 0.77 1.34 1.05 0.67 0.00 0.52 0.45
0.27-0.97 0.14-O.99 0.01-3.18 0.15-2.18 0.01-1.17 0.00-4.42 0.02-4.64 0.00-6.89 0.15-2.15 0.35-1.61 0.37-2.64 0.17-2.39 0.01-1.81 0.73-3.77 0.21-1.97 0.58-2.64 0.22-3.06 0.08-2.41 0.00-5.78 0.06-.88 0.01-2.49
0.03 0.04 0.99 0.80 0.10 0.99 0.97 0.98 0.80 0.35 0.63 0.90 0.39 0.11 0.82 0.41 0.77 0.99 0.99 0.60 0.73
• Data on vital status and cancer morbWty as of January 1,1990 (n - 677; 8,709.9 persoo-years of observation), t ICD-7, International Classification ol Diseases; Seventh Revision; Obs, observed nunber of cases; Exp, expected number of cases; SMR, standardzed morbkSty ratio; Cl, confidence interval.
the southern health care region of Sweden). There was a significantly lower incidence of breast cancer than expected—11 tumors observed versus 20.31 expected (SMR = 0.54, p < 0.03). A significantly lower incidence than expected was also seen for tumors of the female genital organs—5 observed versus 11.84 expected (SMR = 0.42, p < 0.04). Low standardized morbidity ratios were noted for tumors of the cervix, uterine body, ovaries, vulva, and vagina. Low standardized morbidity ratios were also seen for lymphomas, leukemias, and pancreatic carcinomas. There was a higher incidence of tumors of the respiratory tract (SMR = 1.83, p < 0.11) than expected. Imposing a latency time (60 months) in the analysis gave similar standardized morbidity ratios as before for total cancer incidence (46 observed cases vs. 66.1 expected; SMR = 0.70), breast cancer incidence (5 observed cases vs. 14.3 cases expected;
SMR = 0.35), and tumors of the female genital tract (4 cases observed vs. 8.1 cases expected; SMR = 0.50). The patients were divided into two groups: those who sustained a fracture before 50 years of age and those who sustained a fracture at or after 50 years of age (table 3). As table 3 shows, reduced cancer morbidity was seen in both age groups, although it was significant only for the older patients. Overall mortality and cause-specific mortality are presented in table 4. Total mortality was lower than expected: 146 women died versus 191.69 expected (standardized mortality ratio (SMR) = 0.76, p < 0.001). There was a reduction in the death rate from malignant tumors (30 observed vs. 40.41 expected; SMR = 0.74, p < 0.09) and a reduction in mortality from circulatory disease (79 observed vs. I l l .49 expected; SMR = 0.71, p < 0.002). Significantly lower mortality than expected was noted for both
Downloaded from https://academic.oup.com/aje/article-abstract/136/4/422/164777 by Columbia University user on 17 January 2019
140-209
Cancer in Women with Forearm Fractures
425
TABLE 3. Cancer morbidity in a prospective cohort of women who had distal forearm fractures in 1974 or 197S, by age (
Vo) 136, No. 4 Printed in U.S.A.
A BRIEF ORIGINAL CONTRIBUTION
Hakan Olsson1 and Gunnar Hagglund2
At the Department of Orthopedics of University Hospital, Lund, Sweden, cancer incidence and overall and cause-specific mortality were studied in a prospective cohort of 677 women who had experienced a fracture of the distal forearm in 1974 or 1975. The women were followed through population-based population, cancer, and death registries in southern Sweden. There was a significant reduction in overall cancer incidence (66 cases observed vs. 90.21 expected), breast cancer incidence (11 cases observed vs. 20.31 cases expected), and incidence of tumors of the female genital tract (5 cases observed vs. 11.84 expected). Overall mortality was also reduced (146 cases observed vs. 191.69 cases expected), and rates of death from circulatory disease (79 cases observed vs. 111.49 expected) and malignant tumors (30 cases observed vs. 40.41 expected) were both lower than expected. Am J Epidemiol 1992;136:422-7. breast neoplasms; cohort studies; estrogens; fractures; morbidity; mortality; neoplasms; osteoporosis
consequences of estrogen deficiency, such as peri-/postmenopausal symptoms, fragility fractures, and accelerated cardiovascular disease (5-7). The overall benefit would include increased life expectancy. It has been argued that the possible adverse effects of estrogen supplementation—breast and endometrial carcinoma—would not change the overall positive effect of such supplementation (8). While the association between endometrial cancer and estrogen supplementation is established, the question of whether estrogen replacement therapy increases the risk of breast cancer is still controversial. A recent meta-analysis indicates, however, that the association between long-term estrogen supplementation and breast cancer is real (9). We investigated cancer morbidity and total and cause-specific mortality in a cohort of women with distal forearm fractures in order to gain insight into the total picture of morbidity and mortality in women suffering from osteoporosis. Since forearm fractures generally strike women somewhat earlier
Exposure to endogenous and exogenous estrogens has been implicated as a cause of breast cancer and endometriaJ carcinoma (1-4). It is therefore reasonable to hypothesize that medical conditions leading to or indicating estrogen deficiency would at least partially protect against these types of tumors. Such conditions would include early menopause, body leanness (postmenopausal), and signs of accelerated bone loss (osteoporosis). It has been suggested that estrogen substitution peri- and postmenopausally could prevent some of the medical
Received for publication September 3, 1991, and in final form April 6, 1992. Abbreviation: SMR, standardized morbidity/mortality ratio. 1 Department of Oncology, University Hospital, Lund S-221 85, Sweden. 1 Department of Orthopedics, University Hospital, Lund S-221 85, Sweden. Reprint requests to Dr. Hakan CHsson, Department of Oncology, University Hospital, Lund S-221 85, Sweden. This study was supported by grants from the Medical Faculty of Lund University. 422
Downloaded from https://academic.oup.com/aje/article-abstract/136/4/422/164777 by Columbia University user on 17 January 2019
Reduced Cancer Morbidity and Mortality in a Prospective Cohort of Women with Distal Forearm Fractures
Cancer in Women with Forearm Fractures
MATERIALS AND METHODS
Through computerized diagnostic data bases, we identified all women (n = 677) with a first distal forearm fracture seen in the Orthopedic Department of University Hospital, Lund, Sweden, in 1974-1975. The hospital serves the Health District of Lund and has population-based recruitment. The studied population represents all patients who sustained a distal forearm fracture during the given time period in this geographic area. The data base containing the orthopedic diagnoses in 1974-1975 was fully computerized. The median age of the women was 62 years (range, 23-88 years). At the time of the fracture, 24 women were younger than age 30 years, 108 were aged 30-50 years, 378 were aged 50-70 years, and 167 were over age 70 years. By means of the unique Swedish 10-digit personal identification number, we followed the vital status and cancer incidence of these patients from 1974 onward in the population-based Census Registry, Cause of Death Registry, and Swedish Cancer Registry (the South Swedish Regional Tumour Registry and the National Swedish Tumour Registry). Each individual could have had more than one tumor registered. Observed versus expected numbers of cases were then calculated for various tumor types/causes of death usingreferencedata from the southern health care region. The expected morbidity and mortality figures were derived from the same population of women from which the fracture cases were derived. Expected cancer incidence was
calculated using data from the South Swedish Regional Tumour Registry. In the calculations, we used the person-years method, classifying the individuals into 5-year age groups and using a single calendar year as the unit cell size. Vital status was determined up to January 1, 1990. The vital status of the women is presented in table 1 as of January 1, 1987. Five subjects were lost to follow-up because of insufficient information on identity, and three patients were lost because of emigration. Cause-specific standardized morbidity/ mortality ratios and their 95 percent confidence intervals were calculated. The/? values were calculated by use of the Poisson distribution, or the chi-square distribution if the expected values were greater than 10. The term "significant" refers to p < 0.05. All tests were two-tailed. Data on expected cancer morbidity in the health care region were available and updated as of January 1, 1990. Data on expected cancer mortality were available and updated to January 1, 1987. For assessment of cancer morbidity, diseases were grouped according to the International Classification of Diseases, Seventh Revision, and for the mortality analysis, diseases were grouped according to the International Classification of Diseases, Ninth Revision. RESULTS
Table 2 shows cancer morbidity as of January 1, 1990, for various tumor types. During the follow-up period, 66 women developed malignant tumors, compared with an expected incidence of 90.21 (standardized morbidity ratio (SMR) = 0.73, p < 0.01, in comparison with the expected incidence in TABLE 1. Vital status of a cohort of women with distal forearm fractures, South Sweden, as of January 1, 1987 Vital status
No.
Living Deceased Emigrated from Sweden Unknown (insufficient identity)
523 146 3 65
77.3 21.6 0.4 0.7
Downloaded from https://academic.oup.com/aje/article-abstract/136/4/422/164777 by Columbia University user on 17 January 2019
than hip fractures and (contrary to hip fractures) have not previously been suggested to increase mortality (10), we chose to study this group of women to get a clearer picture of the overall health consequences of osteoporosis. From a 1974 study (conducted by absorptiometry) in our health care region of bone mineral content in the forearms of women with distal forearm fractures (11), it was concluded that this group of women had reduced bone mineral content compared with healthy referents.
423
424
Olsson and Hagglund
TABLE 2. Cancer morbidity in a prospective cohort of women who had distal forearm fractures in 1974 or 1975* ICD-7t code(s)
Type of cancer
ot»t
Expt
SMRt
95% a t
P
All malignant tumors
66
90.21
0.73
0.57-0.94
0.01
170 171-176 171 172 175 176 140-148 150 151 152-153 154 155 157 160-164 180-181 190-191 193 194-195 196-197 200-203 204-209
Breast Female genital organs Cervix uteri Uterine body Ovary Vulva/vagina Mouth/pharynx Esophagus Stomach Small and large bowel Rectal carcinoma Liver and biliary tract Pancreas Respiratory tract Urinary organs Skin tumors Brain Endocrine glands Bone/soft tissue sarcoma Lymphoma Leukemia, potycythemia
11 5 1 1 1 0 1 0 3 8 5 3 1 7 4 8 3 2 0 2 1
20.31 11.84 1.75 4.03 4.78 0.84 1.20 0.54 4.07 9.81 4.42 3.67 3.07 3.83 5.19 5.96 2.87 3.00 0.64 3.85 2.24
0.54 0.42 0.57 0.74 0.21 0.00 0.83 0.00 0.74 0.82 1.13 0.82 0.33 1.83 0.77 1.34 1.05 0.67 0.00 0.52 0.45
0.27-0.97 0.14-O.99 0.01-3.18 0.15-2.18 0.01-1.17 0.00-4.42 0.02-4.64 0.00-6.89 0.15-2.15 0.35-1.61 0.37-2.64 0.17-2.39 0.01-1.81 0.73-3.77 0.21-1.97 0.58-2.64 0.22-3.06 0.08-2.41 0.00-5.78 0.06-.88 0.01-2.49
0.03 0.04 0.99 0.80 0.10 0.99 0.97 0.98 0.80 0.35 0.63 0.90 0.39 0.11 0.82 0.41 0.77 0.99 0.99 0.60 0.73
• Data on vital status and cancer morbWty as of January 1,1990 (n - 677; 8,709.9 persoo-years of observation), t ICD-7, International Classification ol Diseases; Seventh Revision; Obs, observed nunber of cases; Exp, expected number of cases; SMR, standardzed morbkSty ratio; Cl, confidence interval.
the southern health care region of Sweden). There was a significantly lower incidence of breast cancer than expected—11 tumors observed versus 20.31 expected (SMR = 0.54, p < 0.03). A significantly lower incidence than expected was also seen for tumors of the female genital organs—5 observed versus 11.84 expected (SMR = 0.42, p < 0.04). Low standardized morbidity ratios were noted for tumors of the cervix, uterine body, ovaries, vulva, and vagina. Low standardized morbidity ratios were also seen for lymphomas, leukemias, and pancreatic carcinomas. There was a higher incidence of tumors of the respiratory tract (SMR = 1.83, p < 0.11) than expected. Imposing a latency time (60 months) in the analysis gave similar standardized morbidity ratios as before for total cancer incidence (46 observed cases vs. 66.1 expected; SMR = 0.70), breast cancer incidence (5 observed cases vs. 14.3 cases expected;
SMR = 0.35), and tumors of the female genital tract (4 cases observed vs. 8.1 cases expected; SMR = 0.50). The patients were divided into two groups: those who sustained a fracture before 50 years of age and those who sustained a fracture at or after 50 years of age (table 3). As table 3 shows, reduced cancer morbidity was seen in both age groups, although it was significant only for the older patients. Overall mortality and cause-specific mortality are presented in table 4. Total mortality was lower than expected: 146 women died versus 191.69 expected (standardized mortality ratio (SMR) = 0.76, p < 0.001). There was a reduction in the death rate from malignant tumors (30 observed vs. 40.41 expected; SMR = 0.74, p < 0.09) and a reduction in mortality from circulatory disease (79 observed vs. I l l .49 expected; SMR = 0.71, p < 0.002). Significantly lower mortality than expected was noted for both
Downloaded from https://academic.oup.com/aje/article-abstract/136/4/422/164777 by Columbia University user on 17 January 2019
140-209
Cancer in Women with Forearm Fractures
425
TABLE 3. Cancer morbidity in a prospective cohort of women who had distal forearm fractures in 1974 or 197S, by age (
