Mortality from cardiovascular diseases and sudden death in ferroalloy plants

from cardiovascular diseases and sudden in ferroalloy Objectives The aim of this study was to examine mortality from circulatory diseases and sudden death among workers in 12 Norwegian ferroalloy plants. Methods The cohort comprised 14 730 men employed for the first time during 1933-1990 and for at least 6 months. Deaths observed during 1962-1990 were compared with expected figures calculated from national mortality rates. Internal cornparisoils of rates were performed by Poisson regression analysis. Results The overall mortality from cardiovascular diseases was not increased [standardized mortality ratio (SMR) 1.011, but a significantly increased mortality from sudden death (SMR 1.55) and hypertensive disease (SMR 1.37) was observed. Among the ferromanganeselsilicomanganese (FeMnISiMn) furnace worl<ers the sudden death mortality was significantly increased during the employment period (SMR 2.47). In an internal comparison of the sudden death rates, a significant increase of 0.05 in the rate ratio per workyear was observed in this group. The mortality from 3 hypertension-related diseases combined (cerebrovascular, hypertensive, and renal diseases) showed identical positive mortality trends among the ferrosiliconlsilicon-metal (FeSiISi-met) and the FeMnISiMn furnace workers by increasing duration of work. the FeMnISiMn furnace worlters is not likely to be by smolting or alcohol consumption. Associations with work exposures (manganese and possibly with increasing duration of work in both groups of furnace workers be associated with common furnace work

possible with exposure to manganese alloys, as animal studies have shown that the divalent manganese ion (Mn++) may influence cardiac function (3,4).
We have previously reported the main causes of death for a cohort of 14 730 male workers in the Norwegian ferroalloy industry (5). A significantly increased standardized mortality ratio (SMR) of 1.47 for sudden death was found for all men employed at least 3 years. Statistics Norway has formerly shown that workers in smelting plants have the highest mortality index for sudden death among all work categories in Norway (6). There has further been some concern within this industry about a potential association between heat exposure and blood pressure elevation. However, no association between cardiovascular mortality and worlc in this industry has been reported.
Thus the aim of this study was to examine the associations between duration of work in specific departments and lnortality from circulatory system diseases and sudden death within the same cohort (5).

Study setting
The study was conducted among employees in 12 Norwegian electrometallurgical femoalloy plants. Nine of the plants started production before 1930. The remaining three were established in the period 1963-1974. Eight of the 12 plants had primarily produced ferrosilicon (FeSi) or silicon-metal (Si-met) or both. The remaining four had primarily produced ferromanganese (FeMn) and silicomanganese (SiMn).

The exposure variable
Only indirect exposure information from manual employment records was available in addition to sporadic measurements, mainly of dusts. Exposure classification was therefore based on the individual registered periods of work in the actual work areas at the department level (5).
In the 12 plants a total of 81 departments or specialized functions were identified. These were further divided into 16 main groups of departments. The 2 largest groups were the furnace workers in the FeSiISi-met and FeMnISiMn plants. As the other groups of workers were considerably smaller, we also defined 2 groups of bluecollar workers who were not furnace workers. They were the mechanics, electricians, transport workers, raw material workers, packers, and building and construction workers in each of the FeSiISi-met and FeMnISiMn types of plants. Individual information about type of plant, work area (department), and duration of work was thereby used as proxies for true exposure information.
Other groups of employees not included in the 4 main groups were office workers, technicians, engineers, research workers, laboratory workers, workers in the parts and accessories departments, ferrochromium furnace workers, ferrovanadium furnace workers, calcium carbide furnace workers, and electrode paste production workers. These groups of furnace workers and the electrode paste production workers were analyzed separately.

Study population
Men employed for the first time during 1 January 1933 to 31 December 1990 and at least 6 months in 1 of the 12 ferroalloy plants were eligible for the cohort. Altogether 4% (629 persons) of the primary cohort was excluded due to unknown vital status or death before the start of the observation period. The cohort then comprised 14 730 identified men. The variables recorded were name, date of birth, the 5-digit personal identification number, an identification number of the plant, up to 10 employment periods, and smoking habits (from medical records).
Information on individual causes of death and death dates were supplied by linking the cohort to Statistics Norway (the national registery for causes of death). Only the underlying cause of death was supplied and not contributory causes. The causes of death had been coded by Statistics Norway according to the Norwegian version of the International Classification of Diseases (ICD) in use at the time when the deaths occurred. The 7th to 9th editions of the ICD were used. The observation period was restricted to 1 January 1962 to 31 December 1990 due to lack of computerized detailed national cause-specific n~ortality rates before 1962.
The total cohort was observed for 288 886 personyears (table 1). The number of person-years of observation within each of the 4 main subgroups ranged from 45 314 to 72 692, and the number of workers from 2517 to 3695. The distributions by year of birth were almost identical in the 4 subgroups. The distribution by duration of work within the groups showed the shortest median and 95th percentile values for the 2 groups of furnace workers. A total of 3382 men had only worked in other departments and were not included in the 4 main subgroups.

Data analysis
The causes of death used in the analyses are presented together with the corresponding ICD codes, in table 2. The analyses of mortality from hypertension-related diseases comprise deaths from hypertensive disease, cerebrovascular disease and renal diseases (except those of the lower genitourinary tract).   Standardized mortality ratios for specific causes of death were calculated by comparing the observed numbers of deaths with the expected numbers. The expected numbers were calculated from 5-year age-specific rates for each year from 1962 through 1990 for the Norwegiall male population. For all the standardized mortality ratios (SMR) a 95% confidence interval (95% CI) was calculated as presented elsewhere (5). The Epicure program package (7) was used to create event-time tables stratified by both fixed and time-dependent variables. Analyses of mortality i11 the time-related categories of employment status (active and inactive person-time) were performed (8). The active person-time comprises all persontime from the start of follow-up until the end of last employment (including periods off work before that point of time) added by 0.1 year, and inactive person-time is the remaining person-time of observation (5). If a man had worked as both a furnace and a nonfurnace worker he was included in both groups with the actual periods. A distinction between short-and long-term workers was made at 3 years' duration of work.
Internal comparisons of the rates were performed with the Poisson regression analysis with the Epicure program package (7). Confidence intervals (95%) for rate ratios (RR) and P-values for likelihood ratio tests and trend analysis were calculated by the program. The P-value for the trend analysis was calculated as the Pvalue for the likelihood ratio test when the variable was withdrawn from the final model after the categories of the variable had been scored by their mean values. Rural-urban location was included in the regressions due to previous findings (5). As only 18.5% of the dead could be categorized as ever-or never-smokers, analyses of cardiovascular mortality by smoking habits were not performed.

Mortality from ischemic heart disease and sudden death
The mortality from all diseases of the circulatory system was not increased in the total cohort (1 5 18 deaths, SMR 1.01) (table 3). The SMR for ischemic heart disease was 0.98 (999 deaths). A significantly increased mortality from sudden death was found (SMR 1.55, 172 deaths). The sudden deaths were evenly distributed over the observation period. For ischemic heart disease and sudden death combined the SMR was 1.04 (95% CI 0.98-1.10). Table 4 shows a significantly increased mortality from ischemic heart disease among short-term FeSiISimet furnace worlters (SMR 1.35,70 deaths), an excess of sudden deaths among the short-term furnace and nonfurnace workers in the FeSiISi-met plants (SMR 2.36 and 2.45, respectively), and an increased mortality from sudden death among the long-term furnace workers both in the FeSiISi-met plants (SMR 1.70) and in the FeMnI SiMn plants (SMR 2.16).
Further analyses of ~nortality from sudden death in 4 categories of duration of FeMnISiMn furnace work (< 3, 3-9, 10-19, 2 2 0 years) showed SMR values of 1.22 (95% CI 0.56-2.32), 1.61 (95% CI 0.77-2.96), 2.68 (95% CI 1.43-4.58), and 2.39 (95% CI 1.09-4.54), respectively (not shown). A Poisson regression analysis of mortality from sudden death adjusted for age, calendar time, duratio~l of follow-up, and rural-urban location showed a significant increase in rate ratio of 0.050 per workyear (P = 0.01) for the FeMnISiMn furnace workers and a ~lonsig~lificantly increasing rate ratio of 0.019 (P = 0.35) per worltyear for the nonfur~~ace workers in the same plants. Similar analyses for both groups of worlters in the FeSiISi-met plants showed nonsignificantly decreasing rate ratios by increasing duration of work. Table 5 shows that the mortality from sudden death during active person-time was significantly increased for the FeMnISiMn furnace workers (SMR 2.47, 12 deaths) but not for any of the 3 other groups. Of the 12 sudden deaths in the active person-time category 6 occurred during 1962-1969,3 during 1970-1 979, and the 3 last deaths during 1980-1984. Table 6 presents the analyses of mortality from ischemic heart disease and sudden death combined in the   and inactive person-time was found regarding the proportions of sudden deaths. Analyses of mortality in the groups of ferrochromium furnace workers (N = 522), calcium carbide furnace workers (N = 257), ferrovanadium furnace workers (N=66), and electrode paste production workers (N = 342) did not show significantly elevated mortality from any of the studied causes of death (not shown).

Mortality from hypertension-related diseases
In the total cohort the mortality from hypertensive disease was slightly increased (SMR 1.37,51 deaths), while the mortality rates for cerebrovascular and renal diseases were as expected (table 3). Among the long-term workers an increased mortality from cerebrovascular disease was found for the FeMnISiMn furnace workers (SMR 1.39, 52 deaths), as was an increase in hypertensive disease among the nonfurnace workers in the FeMnISiMn plants (SMR 2.18, 11 deaths) (table 4). Stratified analyses of mortality from these 3 hypertension-related causes of death combined indicated increasing mortality by duration of work in both groups of furnace workers (not shown). These associations were further investigated with Poisson regression analyses (table 7). For the com-bined group of FeSiISi-met and FeMnISiMn furnace workers a significantly increasing rate ratio of 0.020 per workyear (P = 0.05) was observed. The corresponding result for the 2 groups of nonfurnace workers was a significantly decreasing rate ratio of 0.022 per workyear (P = 0.01).

Discussion
The main findings from this study were the increased mortality from sudden death among the FeMnISiMn furnace workers and the association between mortality from hypertension-related diseases and the duration of furnace work, both for the FeSiISi-met and the FeMnISiMn workers. Previous studies of cause-specific mortality within the ferroalloy industry have only concerned workers in 2 ferrochromium producing plants (9,10). The main risk factors for coronary heart disease and sudden cardiac death are very similar (1 1-14). The precipitating factors for sudden cardiac deathpossibly as the first manifestation of coronary heart diseaseare less clear (13). In this study the sudden deaths were not only likely to include cardiac deaths, but also other natural causes of sudden death (pulmonary embolism, cerebrovascular disease, etc). However, it has been assumed that all sudden deaths without other explanation are attributable to coronary heart disease (1 1). Therefore, deaths from ischemic heart disease and sudden deaths were combined in some of the analyses.
In Norway, the rules for deciding the underlying cause of death have been much the same during the observation period even if the code numbers have varied through the 7th to the 9th ICD editions. Sudden death (moss subita) has only been used if no other relevant disease was known. This usage was also found in a previous Norwegian study (14) in which sudden death was the underlying cause of death for 50 men (10.3 per 100 000 observation years, age-adjusted), while this diagnosis was mentioned on the death certificate also for another 314 men. Hypertension as the underlying cause of death comprises mainly conditions in which hypertension is mentioned on the death certificate together with strictly defined cardiac or renal disease. In the earlier mentioned Norwegian study (14) 18 men (3.6 per I00 000 observation years, age-adjusted) had hypertension as the underlying cause of death, while hypertension was also mentioned for another 108 men.
In this study, the positive association between duration of work at the FeMnISiMn furnaces and sudden death is of particular interest as the mortality from ischemic heart disease and sudden death combined was not increased or positively associated with duration of work. There was also an excess of sudden deaths for the active person-time category (SMR 2.47), for which the mortality from coronary heart disease and sudden death combined was not elevated (SMR 0.81). This finding strongly indicates an association with work exposures. One would expect low SMR values for the active persontime category (8) due to "the healthy worker effect" (ie, an inappropriate cornparison group for active persontime analyses).
Smoking, alcohol abuse, and elevated blood pressure are risk factors for sudden death (1 1-16). In this study, the influence of these factors could however not be evaluated. But it seems unlikely that the consumption of tobacco and alcohol was higher during the employment period than after employment or that these habits should be positively correlated with duration of furnace work. A potential effect on sudden death mortality from hypertension is more difficult to rule out. But it seems unlikely that elevated blood pressure would increase only sudden death mortality and not mortality from ischemic heart disease.
Work exposures to carbon monoxide, heat, and manganese could theoretically increase the mortality from sudden death. Deaths at work due to acute carbon monoxide exposure or heat stroke would be less likely misclassified as sudden deaths if they were suspected of being accidental, which, according to Norwegian legislation, would probably lead to an autopsy, which in turn could provide the real cause of such deaths. A detrimental effect of chronic exposure to carbon monoxide at a low level among persons with subclinical ischemic heart disease cannot however be ruled out. The level of individual exposure to carbon monoxide among FeMnISiMn furnace workers is unknown, but it has probably not been high, as carbon monoxide has been continuously monitored through stationary sampling in the furnace houses with flashlights or alarms at concentrations of 30 or 50 ppm during most of the observation period.
The possibility that manganese exposure is associated with an increased risk of sudden death is speculative, but should be pursued. In animals the divalent, free manganese ion (Mn++) can severely depress cardiac inotropy and chronotropy when present in high concentrations (3). Cardiotoxic effects are also reported to be the ultimate limiting factor when Mn++ is administered intravenously to humans (3). The effects of long-lasting exposures are however less clear. A Mn++, intragastric ratfeeding study over 14 days (0.25 mmollkg daily) resulted in increased myocardial contractility (4). Work exposure to manganese was estimated from 339 personal samples of total dust and total manganese during 1979-1991 in the largest of the FeMnISiMn plants; the results showed decreasing values during this period (figure I). This result is compatible with the observation that no sudden death occurred in the active person-time category among the FeMnISiMn furnace workers after 1984. Because of a lack of data, we were unable to demonstrate a higher manganese exposure for furnace than for nonfurnace workers in the FeMnISiMn plants. But we believe that such a difference exists. This assu~nption is also in agreement with the findings of an Italian study (17).
In regard to the hypertension-related causes of death, we included renal diseases since preliminary analyses indicated that the death diagnoses of hypertensive disease and renal disease might be inaccurately distinguished among employees in rural plants. The identical increase in the rate ratio per workyear in the FeSiISi-met and FeMnISiMn groups of furnace workers and the similar rate ratios when these groups were co~npared suggest that common exposures related to furnace work, rather than exposures to specific metals, may increase the mortality from hypertension-related diseases.
From studies in iron foundries and steel works associations have been reported between elevated blood pressure and exposure to heat or carbon monoxide (1 [8][9][10][11][12][13][14][15][16][17][18][19][20]. However, in a Norwegian study among workers in 1 of the FeSiISi-met plants, no difference in blood pressure between heat-and nonheat-exposed workers was revealed (21). Psychological stress and shift work have been associated with cardiovascular diseases (2). Both groups of furnace workers were shift workers, but so were some of the no~lfurnace workers. Exposure to noise has been shown to increase blood pressure (22). It was not possible to evaluate, however, the influence of noise exposure on our results, but more than one-half of the noise measurements from one plant exceeded 85 dB(A).
As for most historical cohort mortality studies, several weak points may have distorted our results. The lack of real exposure data, the uncertain death diagnoses, and the use of national rates in the analyses for SMR probably attenuated any true exposure-response relationships.
The lack of data concerning possible confounding factors (eg, smoking, alcohol consumption, and employment or exposure data before and after employ~nent in this industry) may both strengthen and attenuate real exposure-response associations. The internal comparisons are independent of external rates, but may be distorted if unequal distributions of confounding factors occur within a group (eg, FeMnISiMn furnace workers). However, it seems unlikely that confounding factors are positively associated with the duration of work.
In summary, excess sudden deaths among FeMnI SiMn furnace workers have been observed. Associations with work exposures (manganese and possibly carbon monoxide and heat) are suspected. The increasing mortality from hypertension-related diseases with increasing duration of furnace work both in FeSiISi-met and FeMn/ SiMn plants may be related to common furnace work conditions (eg, heat, psychosocial stress, shift work, noise, carbon monoxide).