American Journal of Industrial Medicine 17:493-504 (1990)
Cohort Mortality Study of Seattle Fire Fighters: 1945-1 983 Nicholas Heyer, PhD, Noel S. Weiss, MD, DrPH, Paul Demers, MSPH, and Linda Rosenstock, MD, MPH
Fire fighters are known to be occupationally exposed to many toxic substances. However, the limited number of previous studies has not demonstrated any consistent excess mortality from diseases of a priori concern, such as lung cancer, non-malignant respiratory disease, and cardiovascular disease. We studied 2,289 Seattle fire fighters from 1945 through 1983, and observed 383 deaths. Excess mortality from leukemia (SMR = 503, n = 3) and multiple myeloma (SMR = 989, n = 2 ) was observed among fire fighters with 30 years or more fire combat duty. Lung cancer mortality was elevated (SMR = 177, n = 18) among fire fighters 65 years old or older. We also analyzed the data by considering fire fighters at risk only after 30 years from first exposure. In this analysis, a trend of increasing risk with increasing exposure was observed for diseases of the circulatory system. For this cause of death, fire fighters with 30 years or more fire combat duty had a relative risk of 1.84 compared to those with less than 15 years of fire combat duty. Key words: occupational disease, occupational cancer, occupational cardiovascular disease, fire smoke, healthy worker effect
INTRODUCTION
In the last 10 years, national and regional associations of fire fighters have become increasingly concerned about the toxic properties of fire smoke. Until the 1970s there was little adequate respiratory protection available to protect fire fighters from the fire environment. Even today, limitations on weight and bulk make respiratory protection for fire fighters a difficult problem. The fumes and gases generated from a fire contain numerous toxic substances, including known and suspected carcinogens, and respiratory irritants. Fire smoke is a complex and variable mixture that may include asbestos fibers, polyaromatic and halogenated hydrocarbons, heavy metals (e.g., cadmium, lead, zinc), acrylonitrile, ammonia, benzene, carbon monoxide, cyanides, formaldehyde, isocyanates, sulfides, styrene, and toluene [Kimmerle, 1973; Terril et al., 1978; Treitman et al., Departments of Environmental Health (N.H., P.D., L.R.) and Epidemiology (N.H., N.S. W., P.D.), School of Public Health and Community Medicine, and Department of Medicine, School of Medicine (L.R.), University of Washington, Seattle. Address reprint requests to Nicholas Heyer, Department of Environmental Health SC-34, University of Washington, Seattle, WA 98195. Accepted for publication November 16, 1989.
0 1990 Wiley-Liss, Inc.
494
Heyer et al.
1980; Birky and Clarke, 1981; Hartzell et al., 1983; National Research Council, 1986; Brandt-Rauf et al., 19881. While carbon monoxide, hydrogen cyanide, and hydrogen chloride have been identified as the primary causes of acute fire deaths [Sammons and Coleman, 1974; Dyer and Esch, 1976; Levine and Radford, 19781, information available on the chronic toxicity of fire smoke is sparse and inconsistent. METHODS Cohort Enumeration and Definition
The present cohort study included all male fire fighters (N = 2,289) who were employed by the Seattle Fire Department on or after January 1, 1945, and who had served for at least 1 year prior to January 1, 1980. The cohort was followed and vital status determined through December 31, 1983. A total of 52,914 person-years of observation was accumulated. The cohort was enumerated from files maintained by the Seattle Fire Department. No information was available on race or social security number. Female fire fighters were few, and thus were excluded. Additional information on the cohort was available through files maintained by the Fire Pension Board. Follow-Up and Cause of Death Determination
Follow-up was conducted by reviewing the pension status of each cohort member, and by searching the death records of Washington State and the National Death Index (NDI). A search of Washington State drivers’ license records was also conducted. A vital status of “alive” was assigned to fire fighters who: a) were currently active in the department; b) were currently receiving a pension; or c) had renewed their driver’s license after January 1, 1979, and had not been identified as having died. A vital status of “deceased” was assigned to fire fighters: a) whose spouse was receiving a pension; b) for whom death benefits had been paid; or c) who had matched with one of the death record searches. Additional information on vital status was provided through an extensive network of former fighters. However, the information provided by this network was not used to assign a vital status of deceased unless corroborated by a death certificate. Death certificates were obtained either from the Fire Pension Board or by direct request to the state health departments. Cause of death was coded by a trained nosologist to the ICD classification current at the time of death. Each fire fighter for whom we could not determine vital status was considered lost to follow-up. These men contributed person-years until either the date last employed or the date of last renewal of their driver’s license, whichever was later. Exposure Assignment
Exposure was defined as the time spent working in assignments for which there was the possibility of active fire combat duty (“active” assignment). Assignments meeting this definition were determined by veteran fire fighters working with the Department. All other assignments, such as dispatch duty, were categorized as “ad-
Seattle Fire Fighter Mortality
495
ministrative. ” All “active” assignments, including assignments prior to 1945, were included in the exposure measurement. Statistical Analysis
Standardized mortality ratios (SMRs) were computed by using the National Institute for Occupational Safety and Health (NIOSH) modified life table analysis program [Waxweiler et al., 19831. Comparison rates for U.S. white males were used because the Fire Department reported that the large majority of Seattle fire fighters had been white. The SMRs are reported as the number of observed deaths divided by the number of expected deaths multiplied by 100. Changes in the SMR with the time-related variables of age, years of “active” assignment, and time since first exposure were evaluated. Confidence intervals for the SMRs were calculated according to the methods described by Breslow and Day [1987]. Comparisons between SMRs may be susceptible to bias introduced by large differences in the age structures between the two groups being compared [Breslow and Day, 19871. To allow comparisons between duration-of-employment-subgroups with a minimum of such bias, additional analyses were conducted on a subcohort of fire fighters who survived 30 years after first exposure (the same time interval as the longest exposure duration category). In the subcohort analyses, fire fighters did not start accumulating person-years at risk until 30 years or more had passed since first exposure. Thus, no fire fighters hired after December 31, 1953, could be included in this subcohort, and their exposure profiles may be somewhat different when compared to the entire cohort. The subcohort analyses included the calculation of SMRs by exposure subgroups, and an estimation of relative risks using the lowest exposure category as an internal comparison group. Relative risks were expressed as ratios of the two SMRs. Confidence intervals for the relative risks and Poisson tests for trend among the exposure duration subgroups were calculated according to the methods described by Breslow and Day [ 19871. RESULTS
At the end of the study period, 1,801 (78.7%) of the cohort members were known to be alive, 383 (16.7%) were known to be deceased, and 105 (4.6%) were lost to follow-up. Of those identified as deceased, cause-of-death information was missing for 6 (1.5%). Table I presents cause-specific SMRs for Seattle fire fighters. The all-causes SMR was 76. Less than expected numbers of deaths were observed for cancer (all sites combined) and for all major non-malignant disease categories. Greater numbers of deaths than expected were observed for leukemia (SMR= 173) and other lymphatic/hematopoietic cancers (SMR = 225). However, there were relatively small numbers of deaths due to these causes, and thus there are wide confidence intervals around the SMRs. Tables II-IV present cause-specific mortality and SMRs by age, time since first employment, and duration of employment categories for the entire fire fighter cohort. Table V presents cause-specific mortality and SMRs by duration of employment for the subcohort of fire fighters who survived 30 years after first exposure. Table VI
496
Heyer et al.
TABLE I. Cause-Specific Expected and Observed Deaths and Standardized Mortality Ratios Cause All deaths All cancer Digestive cancer Esophageal cancer Stomach cancer Intcstinal canccr Rectal cancer Respiratory cancer Lung canccr Braininervous system cancer Benigniunspecified, brain Lymphaticihematopoietic cancer Leukemia and aleukeniia Other lymphaticihematopoietic cancers Non-cancer deaths Nervous system disease Circulatory system disease Arteriosclerotic disease Respiratory system disease Acute upper airways Bronchitis Digestive system disease Injury Suicide Homicide
-Observed 383 92 29 1 6 7 2 32 29 3 2 12 7 3
Expected
28 172 133 20 2 0 8 28 5 1
SMR
498.78 95.70 27.34 2.22 5.29 8.82 3.04 31.66 29.88 3.14 0.92 9.46 4.03 1.33
76 96 106 44 1 I3 79 65 101 97 95 218 126 173 225
34.10 217.94 176.87 27.10 0.07 1.60 18.45 39.56 13.20 4.59
82 78 75 73 3,003 0 43 70 37 21
95% C.I. 69-85 77-1 18 71-152 1-250 41-247 32-164 8-237 69-143 65-139 20-279 26-789 65 -222 70-358 47-660 55-1 19 68 -92 63-89 45-1 14 364- 10,841 0-229 19-85 47-102 12-89 1-121
TABLE 11. Observed Deaths From Selected Causes and Standardized Mortality Ratios, by Age Age
Cohort Mortality Study of Seattle Fire Fighters: 1945-1 983 Nicholas Heyer, PhD, Noel S. Weiss, MD, DrPH, Paul Demers, MSPH, and Linda Rosenstock, MD, MPH
Fire fighters are known to be occupationally exposed to many toxic substances. However, the limited number of previous studies has not demonstrated any consistent excess mortality from diseases of a priori concern, such as lung cancer, non-malignant respiratory disease, and cardiovascular disease. We studied 2,289 Seattle fire fighters from 1945 through 1983, and observed 383 deaths. Excess mortality from leukemia (SMR = 503, n = 3) and multiple myeloma (SMR = 989, n = 2 ) was observed among fire fighters with 30 years or more fire combat duty. Lung cancer mortality was elevated (SMR = 177, n = 18) among fire fighters 65 years old or older. We also analyzed the data by considering fire fighters at risk only after 30 years from first exposure. In this analysis, a trend of increasing risk with increasing exposure was observed for diseases of the circulatory system. For this cause of death, fire fighters with 30 years or more fire combat duty had a relative risk of 1.84 compared to those with less than 15 years of fire combat duty. Key words: occupational disease, occupational cancer, occupational cardiovascular disease, fire smoke, healthy worker effect
INTRODUCTION
In the last 10 years, national and regional associations of fire fighters have become increasingly concerned about the toxic properties of fire smoke. Until the 1970s there was little adequate respiratory protection available to protect fire fighters from the fire environment. Even today, limitations on weight and bulk make respiratory protection for fire fighters a difficult problem. The fumes and gases generated from a fire contain numerous toxic substances, including known and suspected carcinogens, and respiratory irritants. Fire smoke is a complex and variable mixture that may include asbestos fibers, polyaromatic and halogenated hydrocarbons, heavy metals (e.g., cadmium, lead, zinc), acrylonitrile, ammonia, benzene, carbon monoxide, cyanides, formaldehyde, isocyanates, sulfides, styrene, and toluene [Kimmerle, 1973; Terril et al., 1978; Treitman et al., Departments of Environmental Health (N.H., P.D., L.R.) and Epidemiology (N.H., N.S. W., P.D.), School of Public Health and Community Medicine, and Department of Medicine, School of Medicine (L.R.), University of Washington, Seattle. Address reprint requests to Nicholas Heyer, Department of Environmental Health SC-34, University of Washington, Seattle, WA 98195. Accepted for publication November 16, 1989.
0 1990 Wiley-Liss, Inc.
494
Heyer et al.
1980; Birky and Clarke, 1981; Hartzell et al., 1983; National Research Council, 1986; Brandt-Rauf et al., 19881. While carbon monoxide, hydrogen cyanide, and hydrogen chloride have been identified as the primary causes of acute fire deaths [Sammons and Coleman, 1974; Dyer and Esch, 1976; Levine and Radford, 19781, information available on the chronic toxicity of fire smoke is sparse and inconsistent. METHODS Cohort Enumeration and Definition
The present cohort study included all male fire fighters (N = 2,289) who were employed by the Seattle Fire Department on or after January 1, 1945, and who had served for at least 1 year prior to January 1, 1980. The cohort was followed and vital status determined through December 31, 1983. A total of 52,914 person-years of observation was accumulated. The cohort was enumerated from files maintained by the Seattle Fire Department. No information was available on race or social security number. Female fire fighters were few, and thus were excluded. Additional information on the cohort was available through files maintained by the Fire Pension Board. Follow-Up and Cause of Death Determination
Follow-up was conducted by reviewing the pension status of each cohort member, and by searching the death records of Washington State and the National Death Index (NDI). A search of Washington State drivers’ license records was also conducted. A vital status of “alive” was assigned to fire fighters who: a) were currently active in the department; b) were currently receiving a pension; or c) had renewed their driver’s license after January 1, 1979, and had not been identified as having died. A vital status of “deceased” was assigned to fire fighters: a) whose spouse was receiving a pension; b) for whom death benefits had been paid; or c) who had matched with one of the death record searches. Additional information on vital status was provided through an extensive network of former fighters. However, the information provided by this network was not used to assign a vital status of deceased unless corroborated by a death certificate. Death certificates were obtained either from the Fire Pension Board or by direct request to the state health departments. Cause of death was coded by a trained nosologist to the ICD classification current at the time of death. Each fire fighter for whom we could not determine vital status was considered lost to follow-up. These men contributed person-years until either the date last employed or the date of last renewal of their driver’s license, whichever was later. Exposure Assignment
Exposure was defined as the time spent working in assignments for which there was the possibility of active fire combat duty (“active” assignment). Assignments meeting this definition were determined by veteran fire fighters working with the Department. All other assignments, such as dispatch duty, were categorized as “ad-
Seattle Fire Fighter Mortality
495
ministrative. ” All “active” assignments, including assignments prior to 1945, were included in the exposure measurement. Statistical Analysis
Standardized mortality ratios (SMRs) were computed by using the National Institute for Occupational Safety and Health (NIOSH) modified life table analysis program [Waxweiler et al., 19831. Comparison rates for U.S. white males were used because the Fire Department reported that the large majority of Seattle fire fighters had been white. The SMRs are reported as the number of observed deaths divided by the number of expected deaths multiplied by 100. Changes in the SMR with the time-related variables of age, years of “active” assignment, and time since first exposure were evaluated. Confidence intervals for the SMRs were calculated according to the methods described by Breslow and Day [1987]. Comparisons between SMRs may be susceptible to bias introduced by large differences in the age structures between the two groups being compared [Breslow and Day, 19871. To allow comparisons between duration-of-employment-subgroups with a minimum of such bias, additional analyses were conducted on a subcohort of fire fighters who survived 30 years after first exposure (the same time interval as the longest exposure duration category). In the subcohort analyses, fire fighters did not start accumulating person-years at risk until 30 years or more had passed since first exposure. Thus, no fire fighters hired after December 31, 1953, could be included in this subcohort, and their exposure profiles may be somewhat different when compared to the entire cohort. The subcohort analyses included the calculation of SMRs by exposure subgroups, and an estimation of relative risks using the lowest exposure category as an internal comparison group. Relative risks were expressed as ratios of the two SMRs. Confidence intervals for the relative risks and Poisson tests for trend among the exposure duration subgroups were calculated according to the methods described by Breslow and Day [ 19871. RESULTS
At the end of the study period, 1,801 (78.7%) of the cohort members were known to be alive, 383 (16.7%) were known to be deceased, and 105 (4.6%) were lost to follow-up. Of those identified as deceased, cause-of-death information was missing for 6 (1.5%). Table I presents cause-specific SMRs for Seattle fire fighters. The all-causes SMR was 76. Less than expected numbers of deaths were observed for cancer (all sites combined) and for all major non-malignant disease categories. Greater numbers of deaths than expected were observed for leukemia (SMR= 173) and other lymphatic/hematopoietic cancers (SMR = 225). However, there were relatively small numbers of deaths due to these causes, and thus there are wide confidence intervals around the SMRs. Tables II-IV present cause-specific mortality and SMRs by age, time since first employment, and duration of employment categories for the entire fire fighter cohort. Table V presents cause-specific mortality and SMRs by duration of employment for the subcohort of fire fighters who survived 30 years after first exposure. Table VI
496
Heyer et al.
TABLE I. Cause-Specific Expected and Observed Deaths and Standardized Mortality Ratios Cause All deaths All cancer Digestive cancer Esophageal cancer Stomach cancer Intcstinal canccr Rectal cancer Respiratory cancer Lung canccr Braininervous system cancer Benigniunspecified, brain Lymphaticihematopoietic cancer Leukemia and aleukeniia Other lymphaticihematopoietic cancers Non-cancer deaths Nervous system disease Circulatory system disease Arteriosclerotic disease Respiratory system disease Acute upper airways Bronchitis Digestive system disease Injury Suicide Homicide
-Observed 383 92 29 1 6 7 2 32 29 3 2 12 7 3
Expected
28 172 133 20 2 0 8 28 5 1
SMR
498.78 95.70 27.34 2.22 5.29 8.82 3.04 31.66 29.88 3.14 0.92 9.46 4.03 1.33
76 96 106 44 1 I3 79 65 101 97 95 218 126 173 225
34.10 217.94 176.87 27.10 0.07 1.60 18.45 39.56 13.20 4.59
82 78 75 73 3,003 0 43 70 37 21
95% C.I. 69-85 77-1 18 71-152 1-250 41-247 32-164 8-237 69-143 65-139 20-279 26-789 65 -222 70-358 47-660 55-1 19 68 -92 63-89 45-1 14 364- 10,841 0-229 19-85 47-102 12-89 1-121
TABLE 11. Observed Deaths From Selected Causes and Standardized Mortality Ratios, by Age Age
