Asbestosis and Mesothelioma Among British
Asbestos Workers (1971–2005)
Anne-Helen Harding, PhD1 and Andrew J Darnton,MSc2
Background Ascertainment of asbestosis and mesothelioma from underlying cause of death underestimates the burden of these diseases. The aims of this study were to estimate the true frequency of asbestosis and mesothelioma among asbestos workers in Great Britain (GB), and to identify factors associated with the risk of death with these diseases. Methods The GB Asbestos Survey was established in 1971 to monitor long-term health outcomes among workers covered by regulations to control asbestos at work. Asbestosis and mesothelioma cases were defined by multiple cause of death, and were ascertained by identifying asbestos workers on the GB Asbestosis and Mesothelioma Registers. Standardized mortality ratios (SMRs) were calculated; the risks of asbestosis and mesothelioma were modeled with Poisson regression analysis. Deaths to the end of 2005 were included.
Results There were 15,557 deaths between 1971 and 2005 among the 98,912 workers. Altogether 477 asbestosis and 649 mesothelioma cases were identified. The SMR for all causes was 1.42, for asbestosis 51.3, and for mesothelioma 13.5. In multiply adjusted analysis, age, sex, job, and birth cohort were significantly associated with asbestosis and mesothelioma. For asbestosis year of first exposure, and for mesothelioma latency, were also statistically significant.
Conclusions The asbestos workers experienced high mortality from all causes, asbestosis, and mesothelioma. There was some evidence that the risk of asbestosis and mesothelioma was lower in later birth cohorts and among those first occupationally exposed to asbestos more recently. Due to the long latency of both diseases, further follow-up is required to confirm these trends. Am. J. Ind. Med. 53:1070–1080 2010.ß2010 Crown copyright.{
KEY WORDS: GB Asbestos Survey; asbestosis; mesothelioma; register
INTRODUCTION
Mesothelioma and asbestosis are major asbestos-related diseases. However, the underlying cause of death recorded on
death certificates tends to underestimate the number of cases. Asbestosis is a chronic disease and many who die whilst suffering from it have a different underlying cause of death. There was no separate cause of death code for mesothelioma before the tenth revision of the International Classification of Diseases (ICD-10), which was implemented in 2001 in England and Wales. Most pleural cancer deaths recorded under ICD-9 were in fact mesotheliomas but many mesotheliomas were coded to ‘‘malignant neoplasm, site unspecified’’ [Hutchings et al., 1995].
The UK Health and Safety Executive (HSE) maintains an Asbestosis Register and a Mesothelioma Register for Great Britain (GB). Details of the registers are reported elsewhere [McElvenny et al., 2005]. The registers were established in 1967, and record all deaths in GB with the terms ‘‘asbestosis’’ or ‘‘mesothelioma’’ mentioned in the 2010 Crown copyright
1
Mathematical Sciences Unit, Health & Safety Laboratory, Harpur Hill, Buxton, Derbyshire, UK
2Statistics Branch, Health & Safety Executive, Redgrave Court, Bootle, Merseyside, UK
Contract grant sponsor: Health and Safety Executive.
*Correspondence to: Dr. Anne-Helen Harding, Health & Safety Laboratory, Harpur Hill, Buxton, Derbyshire SK17 9JN, UK. E-mail: anne-helen.harding@hsl.gov.uk
{
This article Asbestosis and Mesothelioma Among British Asbestos Workers (1971 ^ 2005) was written by Anne-Helen Harding and Andrew J Darnton of Mathematical Sciences Unit and Statistics Branch. It is published with the permission of the Controller of HMSO and the Queen’s Printer for Scotland.
Accepted 25 February 2010
DOI 10.1002/ajim.20844. Published online 4 May 2010 in Wiley Online Library (wileyonlinelibrary.com).
underlying or associated cause of death fields on the death certificate. Deaths are extracted from national death data using textual searches of the cause of death descriptions, and all death certificates selected are manually checked before inclusion in the registers [Health & Safety Executive, 2009a]. The registers provide a consistent data series of mortality over more than four decades, and a more complete estimate of the number of cases of asbestosis and mesothelioma than the underlying cause of death code alone.
The aim of this study was to estimate the burden of asbestosis and mesothelioma among the asbestos workers participating in the GB Asbestos Survey, by identifying asbestos workers whose deaths were recorded on either the Asbestosis or the Mesothelioma Register, and to identify factors affecting the risk of death with these diseases.
METHODS
The GB Asbestos Survey was established in 1970 in order to monitor long-term morbidity and mortality among asbestos workers. The British Medical Association Research Ethics Committee gave approval for the survey. Details of the study have been reported elsewhere [Hutchings et al., 1995; Harding et al., 2009]. Briefly, the survey began recruiting workers predominantly in asbestos products manufacture in 1971. Initially workers were invited to voluntary medical examinations. Statutory medial examinations were required for all working with insulating materials from 1983, and for all occupationally exposed to asbestos above a certain limit from 1987. Workers attending statutory medicals were also included in the survey population. The survey questionnaire, which collected information on first occupational exposure to asbestos, job type, and smoking history, was completed at the time of each medical examination. Detailed exposure information was not collected.
Follow-Up
Participants were flagged for cancer and death registra-tions with the National Health Service Central Register (NHSCR) and the General Register Office for Scotland (GROS) after their first medical. Information on smoking status and job type were updated at second or subsequent medicals.
Job Categories
Workers were grouped into four job categories accord-ing to the job type they had spent most time in. The categories were manufacturing, stripping/removal, ‘‘other exposed,’’ and insulation work. The ‘‘other exposed’’ category included shipbuilding, construction, and miscellaneous sectors.
Ascertainment of Asbestosis and
Mesothelioma Deaths
Deaths among asbestos workers who had asbestosis or mesothelioma were identified by their presence on the GB Asbestosis or Mesothelioma Registers, respectively. Thus, the deaths were coded as ‘‘multiple cause,’’ which implied that for asbestosis cases, ‘‘asbestosis’’ was in the underlying or an associated cause of death text field. Consequently an individual died with asbestosis, and not necessarily from asbestosis, and for the purposes of this study, an individual listed on the Asbestosis Register was included in the asbestosis category, irrespective of the underlying cause of death. Mesothelioma was defined in a similar manner. Asbestosis and mesothelioma were not mutually exclusive categories since both terms may appear on a death certificate and the death will then appear on both registers.
The Caldicott Guardian gave permission to link the Asbestos Survey database with the Asbestosis and Mesothelioma Registers, in order to identify survey parti-cipants who had died with asbestosis or mesothelioma. Survey participants, who appeared on either register were identified by matching surname, date of birth, and sex. These matches were then checked manually using first name, date of death, underlying cause of death, date of medical examination, and occupation. The period of observation for asbestosis deaths was restricted to 1978–2005, since prior to 1978 Asbestosis Register data were not available in electronic format.
Statistical Analysis
Standardized mortality ratios (SMRs) were calculated by sex-specific 5-year age and calendar periods for all causes of death, asbestosis, and mesothelioma using OCMAP software [Marsh et al., 2004]. National mortality rates obtained from the Office for National Statistics (ONS) and GROS were used to calculate the SMRs for all causes of death. Underlying cause of death among the asbestos workers was used in the calculation of all cause mortality. The HSE provided national rates for deaths on the Asbestosis and Mesothelioma Registers, so that asbestosis and mesothe-lioma SMRs could be calculated. Using the HSE rates ensured that asbestosis and mesothelioma were defined in the same way in the study and the reference populations. Poisson regression analysis was undertaken to examine the combined effect of potential risk factors for asbestosis and mesothe-lioma, using Stata/SE 10.1 [Stata Corp, 2007]. Person years of follow-up started from the date of entry into the Asbestos Survey, and ended on the earlier of death, emigration from GB, or the end of the study period on December 31, 2005. Factors considered in the analysis were sex, age attained, smoking status, job, birth cohort, period of death, year of first occupational exposure to asbestos, age at first exposure,
duration of exposure, time since last exposure, and latent period (years since first occupational exposure to asbestos). High collinearity between variables, such as between the time-related variables, can lead to the regression coefficients of one or both variables changing substantially when fitted in a model simultaneously. Regression coef-ficients were considered to be stable if the association observed with mortality in the simple model adjusted only for age and sex was similar as in the multiply adjusted model. Consequently, the stability of regression coefficients as well as their statistical significance (P 0.05) was taken into account when adding new terms into the final multiply adjusted models. Model goodness-of-fit was tested to ensure that the assumptions of Poisson regression were not violated.
RESULTS
During the study period 1971 to 2005, 99,680 men and women completed 209,670 questionnaires. Altogether there were 98,912 workers included in the mesothelioma analysis and 98,510 in the asbestosis analysis, after excluding workers with incomplete data, workers with an age outside the range of minimum school leaving age to 84 years at the first medical, or workers not traced at the NHSCR or GROS. The median age of participants at entry to the study was 33 years (inter-quartile range 25–44), nearly 5% of participants were female, and over 50% were current smokers at the last medical examination. Overall there were 15,557 deaths from all causes in the cohort. There were 477 deaths with asbestosis recorded on the death certificate, but only 116 of these had asbestosis as the underlying cause. Of 649 deaths, which had mesothelioma recorded on the death certificate, 61% were coded as pleural or peritoneal cancer (ICD-8 and ICD-9), or mesothelioma (ICD-10) in the underlying cause. Table I shows the number of cases and the total number of individuals at risk by each of the potential explanatory variables included in the analysis.
TABLE I. Characteristics of the Asbestos Workers in the Survey
Asbestosis Mesothelioma Deaths Persons at risk Deaths Persons at Risk Total 477 98,510 649 98,912 Men 466 94,021 631 94,403 Women 11 4,489 18 4,509 Smoking statusa Current smoker 303 52,277 321 52,575 Former smokers 144 18,966 210 19,030 Never smoker 24 24,414 102 24,441 Job typea Manufacturing 132 28,286 214 28,578 Stripping/removal 99 51,224 200 51,224 Other 65 12,796 87 12,859 Insulation worker 177 5,165 143 5,212 Birth cohort <1920 111 4,290 85 4,543 1920^ 210 9,057 211 9,160 1930^ 119 12,111 211 12,139 1940 ^ 33 17,787 118 17,799 1950^ 4 23,084 21 23,090 1960^ C 21,270 3 21,270 1970 ^ C 10,911 C 10,911
Year of first exposure
<1930 7 96 3 106 1930^ 55 1,042 42 1,093 1940 ^ 110 2,665 100 2,736 1950^ 162 5,914 224 6,012 1960^ 80 12,423 147 12,534 1970 ^ 53 25,219 93 25,280 1980^ 9 25,011 34 25,011 1990^ 1 26,140 6 26,140
Age at first exposure (years)
<20 188 19,500 255 19,554
20^ 111 38,726 173 38,806
30^ 81 22,424 98 22,522
40 ^ 56 12,337 67 12,426
50^ 41 5,523 56 5,604
Duration of exposure (years)
<10 20 22,709 45 22,831
10 ^ 47 27,333 50 27,436
20^ 90 24,878 135 24,970
30^ 122 14,233 194 14,293
40 ^ 198 9,357 225 9,382
Time since last exposure (years)
<10 249 35,945 334 36,347 10 ^ 170 41,032 225 41,032 20^ 58 20,490 89 20,490 30^ 0 1,043 1 1,043 Latency (years) <10 17 19,358 30 19,468 10 ^ 39 21,436 48 21,536 20^ 68 26,419 118 26,510 30^ 115 18,505 184 18,574 40 ^ 143 8,385 188 8,414 50^ 79 3,498 76 3,501 60^ 16 909 5 909 a
Missing values in this variable.
TABLE I. (Continued) Asbestosis Mesothelioma Deaths Persons at risk Deaths Persons at Risk
Tables II and III summarize the underlying causes recorded on death certificates of the deaths classified as asbestosis or mesothelioma. Of deaths classified as asbes-tosis, the underlying cause on death certificates was most often recorded as MN of the trachea, bronchus, and lung (36%), followed by asbestosis (24%), diseases of the circulatory system (8%), and respiratory disease (excluding asbestosis) (6%). For deaths classified as mesothelioma, which occurred before 2001, the underlying cause was most often recorded as MN of the pleura or peritoneum (51%), followed by cancers of ill-defined, secondary and unspecified sites (32%), and cancers of the trachea, bronchus, and lung (5%). For those, which were coded to ICD-10, 89% were coded as mesothelioma (C45), 2% as respiratory disease excluding asbestosis, and 1% as asbestosis. The impact of the implementation of a specific code for mesothelioma under ICD-10 is also apparent in Table III. Before 2001, 32% of deaths classified as mesothelioma were recorded as cancer of ill-defined, secondary and unspecified sites in the underlying cause; from 2001 there was only one such death among asbestos workers with this underlying cause. Of deaths coded as C45, 60% were recorded as ‘‘mesothelioma, unspecified,’’ 18% as peritoneal
mesothelioma, 16% as pleural mesothelioma, and 6% as mesothelioma of other sites. Figure 1 shows the overlap between deaths from lung cancers and deaths with asbestosis or mesothelioma. Of the 1886 deaths from lung cancer, for the purpose of this analysis 9% were classified as asbestosis, 1% as mesothelioma, and there were three deaths (<1%) with both conditions.
Table IV shows the SMRs for all causes of death, asbestosis, and mesothelioma. Compared with the GB population, the asbestos workers experienced a statistically significant excess of deaths from all causes (SMR 1.42, 95% CI 1.39–1.44), and a higher than expected number of deaths with asbestosis (SMR 51.3, 95% CI 46.8–56.1), or mesothelioma (SMR 13.5, 95% CI 12.5–14.6). Women had higher SMRs than men.
Relative risks for each factor considered in the Poisson regression analyses, adjusted for age and sex, are shown in Table V. Smoking status was strongly associated with asbestosis; former and never smokers had significantly lower relative risks compared to current smokers (RR 0.73, 95% CI 0.60–0.90 and RR 0.23, 95% CI 0.15–0.34, respectively). There was no association between smoking status and the relative risk of mesothelioma. For the other potential risk TABLE II. Underlying Cause of Death by Period of Death, for Deaths on the Asbestosis Register (1978^2005)
Cause of death 1978 ^ 1981 ^ 1986 ^ 1991 ^ 1996 ^ 2001 ^ Total
Certain infectious and parasitic diseases 1 1
MN lip, oral cavity, and pharynx 1 1
MN stomach 1 1 1 1 2 6
MN rectum 1 1
MN retroperitoneum and peritoneum 1 5 5 3 2 16
MN other and ill-defined sites within digestive organs and peritoneum 1 1
MN larynx 1 1
MN trachea, bronchus, and lung 11 37 41 41 21 19 170
MN pleura 5 10 7 6 1 29
MN mesothelioma (C45) 11 11
MN bone, connective, and other soft tissue 1 1
MN bladder 1 1
MN kidney 1 1
MN ill-defined, secondary, and unspecified sites 2 4 8 9 3 1 27
MN lymphatic and hematopoietic tissue 1 1
Benign neoplasms 1 17 18
Neoplasms of uncertain behavior or unspecified nature 1 1 2
Endocrine, nutritional and metabolic diseases, and immunity disorders 1 1
Mental disorders 1 1
Diseases of the circulatory system 1 7 5 11 12 36
Diseases of the respiratory system, excluding asbestosis 3 3 5 5 6 6 28
Asbestosis 1 22 13 20 26 34 116
Diseases of the digestive system 1 1 2 1 5
External causes of morbidity and mortality (accidental poisoning) 2 2
factors analyzed, many of the observed associations were similar for asbestosis and mesothelioma though in some cases larger differences in relative risk were seen for asbestosis. Insulation workers had the highest relative risks compared to manufacturing workers (RR asbestosis 8.22, 95% CI 6.56–10.3; RR mesothelioma 4.03, 95% CI 3.26– 4.99), followed by removal workers (RR asbestosis 1.72, 95% CI 1.32–2.24; RR mesothelioma 1.92, 95% CI 1.58– 2.34). The relative risk of asbestosis was higher for those dying during the 1980s, but there was no association between period of death and mesothelioma risk. There was some evidence that birth cohort was inversely related to mesothelioma risk, but there was a strong inverse relationship with asbestosis risk. Year of first occupational exposure to asbestos and age at first exposure were inversely associated with the risks of asbestosis and mesothelioma. Duration of exposure and latency were positively associated with asbestosis and mesothelioma risk. The risk of asbestosis leveled off at 50þ years’ latency, while the risk of mesothelioma peaked at 40–59 years’ latency. The risk of asbestosis was statistically significantly lower 10 or more years after last exposure to asbestos; compared to <10 years, RR was 0.76 (95% CI 0.62–0.93) for 10–19 years and 0.68 (95% CI 0.51–0.92) for 20–29 years since last TABLE III. Underlying Cause of Death by Period of Diagnosis, for Deaths on the Mesothelioma Register (1971 ^2005)
Cause of death 1971 ^ 1976 ^ 1981 ^ 1986 ^ 1991 ^ 1996 ^ 2001 ^ Total
MN stomach 1 1 1 1 4
MN rectum 1 1
MN liver and intrahepatic bile ducts 1 1
MN retroperitoneum and peritoneum 11 16 20 26 17 90
MN other and ill-defined sites within digestive organs and peritoneum 1 1
MN trachea, bronchus, and lung 5 3 9 7 1 25
MN pleura 2 13 31 25 45 40 156
MN mesothelioma (C45) 152 152
MN bone, connective, and other soft tissue 1 1
MN respiratory and intrathoracic organs 2 1 1 4
MN breast (female) 1 1
MN prostate 1 1
MN ill-defined, secondary and unspecified sites 1 3 11 27 56 57 1 156
MN lymphatic and hematopoietic tissue 1 1 1 3
Benign neoplasms 1 1 2
Neoplasms of uncertain behavior or unspecified nature 1 1
Diseases of the circulatory system 2 1 2 7 1 13
Diseases of the respiratory system, excluding asbestosis 1 1 3 2 2 3 12
Asbestosis 1 1 2 1 5 2 12
Diseases of the digestive system 1 1
Diseases of the musculoskeletal and connective tissue 1 1
Diseases of the genitourinary system 1 1
External causes of morbidity and mortality (accidental poisoning) 1 8 1 10
Total 4 32 68 83 145 147 170 649 Asbestosis 12 820 220 (1%) 12,820 82 % 88 (1%) MN Lung 543 (4%) 1,702 (11%) 15 (<1%) 3 (<1%) 166 (1%) Total deaths = 15,557
Asbestosis: Case listed on the Asbestosis Register Mesothelioma: Case listed on Mesothelioma Register
% of total
Mesothelioma
Mesothelioma: Case listed on Mesothelioma Register MN Lung: Cancers of trachea, bronchus & lung defined by underlying
FIGURE 1.Venn diagram showingthe overlap between lung cancer deaths and deaths with asbestosis or mesothelioma recorded on the death certificate.
exposure. For mesothelioma there was little evidence of a reduction in risk with time since last exposure.
The final regression models are summarized in Table VI. The variables retained made a statistically significant contribution in the presence of the other variables in the model, and had stable regression coefficients. The variables included in the final model for asbestosis were age, sex, smoking status, main job, cohort, and year of first exposure. The risk of asbestosis was 79% lower for those born in the 1960s compared with those born before 1930, and was 87% lower for those first exposed in the 1960s compared with those first occupationally exposed to asbestos before 1930. In the asbestosis model, year of first exposure explained most of the variation in the time-related variables: latency, duration of exposure, and time since last exposure, were not statisti-cally significant, and period and age at first exposure had unstable coefficients, in the presence of year of first exposure. When year of first exposure was not included in the model, there was a highly significant positive association between latency and asbestosis mortality (P < 0.001). However, because of difficulties in interpreting latency for asbestosis as defined for this study, year of first exposure rather than latency was included in the model. The variables included in the mesothelioma model were age, sex, main job, cohort, and latency. Age at first exposure and duration of exposure were statistically significant in the multiply adjusted model but due to collinearity, whereby their associations with mesothe-lioma mortality were reversed in the presence of the other explanatory variables, they were excluded from the model. The effect of cohort was strengthened in the final model: the risk of mesothelioma was 72% lower for those born in the 1960s, and 80% lower for those born in the 1970s, than for those born before 1930.
DISCUSSION
The high rates of mortality from all causes, all malignant neoplasms and specific asbestos-related diseases among this large cohort of British asbestos workers have been
docu-mented elsewhere [Author et al., 2009]. Focussing on mesothelioma and asbestosis, this study highlights more clearly the full extent of these two important asbestos-related diseases. Of the 15,557 deaths from all causes, 649 were classified as mesothelioma: these account for nearly 6% of expected deaths from all causes and represent a 13.5-fold increase over the expected number of mesothelioma deaths based on national rates. Despite this high rate of mortality, mesothelioma deaths among these British asbestos workers account for a small proportion of the 33,000 mesothelioma deaths that occurred during the period 1970–2005 [Health & Safety Executive, 2009b].
Mesothelioma rates were particularly high for some groups of asbestos workers. For example, 7% of the 1,755 deaths that have so far been observed among the 17,799 workers born during the 1940s were due to mesothelioma. Thus, the lifetime risk of these workers is of the same order as that now seen among British men born in the same decade and who worked as carpenters for 10 years before the age of 30 [Peto et al., 2009]. The much smaller numbers exposed compared with those exposed during carpentry and other construction work means deaths among British asbestos manufacturing and licensed workers will make a relatively small contribution to the overall burden of mesothelioma in Britain.
Age, sex, main job, and birth cohort were significant predictors of the risks of mesothelioma and asbestosis in the multiply adjusted Poisson regression analyses. In addition, smoking status and year of first occupational exposure were significantly associated with asbestosis risk, while latent period was associated with mesothelioma risk in the final models. These findings are consistent with those of other studies of workers occupationally exposed to asbestos [Ross and McDonald, 1995; Pira et al., 2005; Hein et al., 2007].
The declining risk of asbestosis and mesothelioma among later birth cohorts, and those first occupationally exposed in more recent times (and for asbestosis only, in later periods of death) suggest that measures to control asbestos exposure among British asbestos workers have led to reduced TABLE IV. Standardized Mortality Ratios for all Causes, Deaths With Asbestosis, and Deaths With Mesothelioma
Cause of death Person years follow-up Deaths observed Deaths expected SMR
All causes 1,780,494 15,557 10,978 1.42 (1.39 ^1.44) Men 1,677,102 14,677 10,412 1.41 (1.39^ 1.43) Women 103,392 880 566 1.55 (1.45^1.66) Asbestosis 1,705,391 477 9.3 51.3 (46.8 ^56.1) Men 1,609,291 456 9.1 50.3 (45.8^ 55.2) Women 96,100 11 0.2 254 (127 ^454) Mesothelioma 1,780,494 649 48 13.5 (12.5^14.6) Men 1,677,102 631 47.4 13.3 (12.3 ^14.4) Women 103,392 18 0.6 30.9 (18.3^48.8)
TABLE V. Poisson Regression Analysis: Relative Risks of Asbestosis and Mesothelioma Adjusted forAge and Sex
Risk factor
Asbestosis Mesothelioma
Relative risk 95% CI LR testa(df) Relative risk 95% CI LR testa(df)
Smoking status 1,040 (6) 1,030 (5) Current smoker 1.00 1.00 Former smokers 0.73 0.60^0.90 1.07 0.90 ^1.28 Never smoker 0.23 0.15 ^0.34 0.88 0.71 ^1.11 Job type 1,292 (7) 1,213 (6) Manufacturing 1.00 1.00 Stripping/removal 1.72 1.32^2.24 1.92 1.58^ 2.34 Other 1.37 1.02^ 1.85 1.13 0.88 ^1.45 Insulation worker 8.22 6.56 ^10.3 4.03 3.26^4.99 Period of death 1,057 (7) 1,066 (6) 1970 ^ 1.00 1.00 1980 ^ 2.88 1.67 ^4.98 1.37 0.90^2.06 1990 ^ 1.48 0.85^ 2.55 1.49 1.00^2.22 2000^ 1.04 0.60^1.83 1.25 0.84 ^1.88 Birth cohort 1,094 (10) 1,084 (9) <1920 1.00 1.00 1920 ^ 0.89 0.71 ^1.13 1.21 0.94^1.56 1930 ^ 0.50 0.38 ^0.67 1.24 0.94^1.62 1940^ 0.18 0.12 ^0.28 1.01 0.73 ^1.40 1950 ^ 0.05 0.02^0.15 0.45 0.26^0.77 1960 ^ C C 0.33 0.10 ^1.14 1970 ^ C C C C
Year of first exposure 1,352 (11) 1,352 (10)
<1930 1.00 1.00 1930 ^ 0.62 0.28^ 1.37 1.08 0.34^ 3.49 1940^ 0.51 0.24 ^1.10 1.02 0.32 ^3.23 1950 ^ 0.40 0.19 ^0.86 1.17 0.37^3.67 1960 ^ 0.13 0.06^0.28 0.49 0.15^ 1.53 1970 ^ 0.07 0.03^0.14 0.23 0.07 ^0.73 1980 ^ 0.03 0.01 ^0.07 0.19 0.06^ 0.62 1990 ^ 0.01 0.001 ^0.09 0.12 0.03^0.48
Age at first exposure (years) 1,206 (8) 1,360 (7)
<20 1.00 1.00
20^ 0.37 0.29 ^0.47 0.42 0.35^0.51
30^ 0.24 0.19 ^0.32 0.22 0.18 ^0.28
40^ 0.16 0.12 ^0.22 0.16 0.12 ^0.20
50^ 0.17 0.12^ 0.24 0.21 0.15 ^0.28
Duration of exposure (years) 1,198 (8) 1,377 (7)
<10 1.00 1.00
10^ 1.29 0.76^ 2.18 0.67 0.45^1.01
20^ 2.17 1.34^3.55 1.75 1.24 ^2.47
30^ 4.06 2.51 ^6.57 3.70 2.64^ 5.19
40^ 8.87 5.46^14.4 6.60 4.65^9.37
Time since last exposure (years) 991 (7) 1,061 (6) 1,061 (6)
<10 1.00 1.00
10^ 0.76 0.62^0.93 0.90 0.76 ^1.08
20^ 0.68 0.51 ^0.92 0.99 0.78^ 1.26
levels of mortality related to these diseases. However, their very long latency means that men and women exposed more recently are still at risk in the future, and it will be many years before the full extent of the risk can be described.
The risk of mesothelioma was found to increase strongly with time since first exposure to asbestos, but the risk did not continue to increase indefinitely and appeared to decline with very long follow-up. This is consistent with the findings of other studies [Berry et al., 2004; McDonald et al., 2006]. However, due to the relatively small number of deaths at the longest latencies (>60 years) these findings cannot be regarded as conclusive.
The concept of a ‘‘latent period,’’ or delay between exposure and the onset of clinically detectable disease, is less straightforward for asbestosis since it is a chronic disease that develops gradually over time. Though it may be detectable on X-rays or by examination of lung tissue, in the early stages there may be no physical symptoms or disability. As with mesothelioma, the time between first exposure and death was referred to as the ‘‘latency’’: in reality the latent period is likely to be a substantial overestimate. We nevertheless found that asbestosis risk was associated with this measure of latency, though interpretation of the relationship was difficult in the context of other highly correlated time-dependent variables.
Identifying cases from the asbestosis and mesothelioma registers provided a better estimate of the frequency of asbestosis and mesothelioma within the cohort. Overall, case ascertainment using underlying cause of death code identified only 24% of the workers with asbestosis, and 61% of workers with mesothelioma as classified by linkage to the asbestosis and mesothelioma registers. However, ascer-tainment of mesothelioma deaths improved substantially after the introduction of ICD-10 with nearly 90% of the deaths in this cohort being recorded as mesothelioma (code C45). An added benefit of the C45 code was that pleural and
peritoneal cancers were no longer included as ‘‘mesothelio-mas.’’ This was particularly significant for peritoneal cancer before the introduction of ICD-10: studies undertaken in England reported that the ICD-8 code (158.9) for peritoneal cancer included only a small proportion of mesotheliomas [Gardner et al., 1985], whereas the ICD-8 code (163.0) for pleural cancer comprised 90% mesotheliomas in men and 70% mesotheliomas in women [Gardner et al., 1982]. A French study estimated that 80% of male deaths, and 48% of female deaths, coded to ICD-9 163 (pleural cancer) were pleural mesotheliomas [Ilg et al., 1998], and in New Zealand 71% of pleural cancers occurring between 1980 and 1996 were recorded as mesothelioma [Kjellstrom and Smartt, 2000]. Although overall case ascertainment improved with C45, specification of the mesothelioma site did not improve. This will impact any studies in which the ability to distinguish between pleural and peritoneal mesotheliomas is important.
The proportion of deaths classified as asbestosis with underlying cause of death recorded as such (26%) was similar to the proportion seen in the national GB Asbestosis Register data during the period [Health & Safety Executive, 2009c]. In the United States 33% of the 25,564 asbestosis deaths, which occurred between 1968 and 2004, identified through multiple cause of death records had asbestosis as the underlying cause of death [Antao et al., 2008]. In a New Zealand study of all men diagnosed with asbestosis between 1974 and 2001, asbestosis was the underlying cause in 17% of deaths with asbestosis during the period 1988–1999, and of the remaining deaths with asbestosis, 63% were cancers of the lung or pleura [Smartt, 2004]. The equivalent statistics for the GB asbestos workers during the period 1978–2000 (21% and 60%, respectively) were remarkably similar.
A major strength of the GB Asbestos Survey was that case ascertainment using the national registers ensured comprehensive coverage of the population, and was unlikely
Latency (years) 1,235 (10) 1,377 (9) <10 1.00 1.00 10^ 1.06 0.60^1.88 0.81 0.51 ^1.28 20^ 1.49 0.87 ^2.55 1.73 1.15 ^2.60 30^ 3.19 1.90^5.36 3.69 2.47 ^5.50 40^ 6.95 4.12^ 11.7 6.89 4.57 ^10.4 50^ 10.7 6.13 ^18.7 8.87 5.61 ^14.0 60^ 11.6 5.61 ^23.8 3.48 1.32^ 9.19 a
Model likelihood ratio chi-squared value and degrees of freedom. Asbestosis was adjusted for linear, quadratic, and cubic terms for age, while mesothelioma was adjusted for linear and quadratic terms for age.
TABLE V. (Continued)
Risk factor
Asbestosis Mesothelioma
to be biased or affected by the medicolegal environment. Studies, which identify cases from occupational compensa-tion schemes, may be subject to these influences [Goldberg et al., 2006; Marrett et al., 2008]. The data collected on asbestos-related cancers tend to be more comprehensive than the data collected on asbestosis. The cancers are identified through cancer registries and national surveillance programs but, with some exceptions, asbestosis surveillance tends to
rely on workers’ compensation claims or underlying cause on death certificates, and estimates from these are subject to over- and under-estimation. Case ascertainment through linkage with the national registers provided an additional benefit: the national rates for deaths recorded on the Asbestosis and Mesothelioma Registers could be used in the calculation of SMRs. In the absence of national rates based on multiple cause of death coded data, this study would TABLE VI. Poisson Regression Analysis: MultiplyAdjusted Relative Risks of Asbestosis and Mesothelioma
Risk factor
Asbestosis Mesothelioma
Relative risk 95% CI Relative risk 95% CI Age (years)
Linear term 4.72 1.40^ 15.9 1.37 1.24 ^1.51
Quadratic term 0.98 0.95^0.99 0.99 0.996^0.998
Cubic term 1.0001 1.00001 ^1.003 Not statistically significant
Sex (female) 0.45 0.24 ^0.82 0.46 0.28^0.73
Smoking status
Current smoker 1.00 Not statistically significant
Former smokers 0.64 0.52^ 0.78 Never smoker 0.24 0.16 ^0.36 Job type Manufacturing 1.00 1.00 Stripping/removal 3.02 2.27 ^4.02 2.12 1.79 ^2.70 Other 1.56 1.15 ^2.10 1.11 0.87 ^1.43 Insulation worker 7.19 5.70 ^9.07 3.34 2.70 ^4.15 Birth cohort <1920 1.00 1.00 1920^ 1.04 0.83^1.37 0.97 0.75 ^1.26 1930^ 0.75 0.57^1.08 0.84 0.63^1.11 1940^ 0.43 0.24 ^0.64 0.61 0.43^0.86 1950^ 0.21 0.04^0.32 0.28 0.16 ^0.48 1960^ C C 0.20 0.06^0.70 1970 ^ C C C C
Year of first exposure
<1930 1.00 Not statistically significant
1930^ 0.52 0.23^1.15 1940^ 0.39 0.18 ^0.86 1950^ 0.32 0.15^0.70 1960^ 0.13 0.06^0.30 1970 ^ 0.07 0.03^0.16 1980^ 0.03 0.01^ 0.09 1990^ 0.02 0.002 ^0.15 Latency (years)
<10 Not statistically significant 1.00
10^ 0.87 0.55 ^1.38 20^ 2.04 1.35 ^3.09 30^ 4.00 2.68^5.98 40^ 6.58 4.36 ^9.92 50^ 8.44 5.33 ^13.4 60^ 3.39 1.28^8.97
have been restricted to internal comparisons for the analyses of asbestosis and mesothelioma.
Although ascertainment of cases through the national registers was unlikely to be biased, the accuracy of the information on death certificates has been the subject of a number of studies. An early study of notifications to the Mesothelioma Register reported that 60% of the 413 notifications in 1967–1968 were histologically con-firmed as mesothelioma while 18% were ‘‘definitely not’’ mesothelioma [Greenberg and Davies, 1974]. Of the 76 ‘‘definitely not,’’ 46% had mesothelioma as the under-lying cause of death on the death certificate. Since 1986, Australia has maintained a Mesothelioma Register; in the subset of cases which underwent a pathology review, 73% of the cases were confirmed as ‘‘definite’’ and 4% as ‘‘definitely not’’ with the remaining 23% ‘‘probable’’ and ‘‘possible’’ cases [Leigh and Driscoll, 2003]. Among the GB asbestos workers, there were 317 deaths during 1996–2005 that matched to the mesothelioma register; 226 (71%) of these were also identified on the regional cancer registries. Thus, 91 (29%) of the deaths on the Mesothelioma Register were not confirmed by cancer incidence data and may have been misclassified. However, there are concerns about the variable quality of data collected by the regional cancer registries in terms of timeliness, completeness, and accuracy [National Statistics, 2001], so a diagnosis of mesothelioma is not ruled out for the unconfirmed notifica-tions. There were an additional 15 incident cases of mesothelioma, identified by the cancer registries and who had died between 1996 and 2004, but were not captured by the Mesothelioma Register.
Similar comparisons between the Asbestosis Register data and pathology/histology data have not been undertaken. Information on study participants who received industrial injuries’ compensation before 1997 was available: of the 363 participants certified as suffering from asbestosis with at least 10% disability, 303 (83%) had died by December 2005, 109 (30%) were listed on the Asbestosis Register, and 33 (30%) of the latter had asbestosis recorded as the underlying cause of death. These statistics highlight the difficulty, which has been reported elsewhere [e.g., Berry, 1981; Mizell et al., 2009], in ascertaining asbestosis cases using death certificate and clinical data rather than pathology/histology. While recognizing the limitations of the Asbestosis Register, and to a lesser extent the Mesothelioma Register, they do provide consistent data series over an extended period.
CONCLUSIONS
The value in maintaining the national asbestosis and mesothelioma registers in order to monitor the occurrence of these significant occupational diseases was highlighted by this study; only one quarter of deaths with asbestosis and
two-thirds of deaths with mesothelioma were identified through the underlying cause of death recorded on death certificates. Potentially important findings from the study were the declining risks of asbestosis and mesothelioma observed for those occupationally exposed in recent times. This trend may reflect improved measures to control exposure to asbestos at work but longer follow-up is required to confirm this.
ACKNOWLEDGMENTS
We would like to thank the staff at the Health and Safety Laboratory and the Health and Safety Executive who, work on the Asbestos Survey. In particular we would like to thank Paul Benson (HSE) for his assistance in accessing data from the Asbestosis and Mesothelioma Registers, and for provid-ing national mortality rates for deaths on the registers. We would also like to thank the staff at the NHSCR, GROS, the occupational physicians, and the asbestos workers for their support.
REFERENCES
Antao VC, Pinheiro GA, Wassell JT. 2008. Asbestosis mortality in the United States: Facts and predictions. Occup Environ Med 66:335– 338.
Berry G. 1981. Mortality of workers certified by pneumoconiosis medical panels as having asbestosis. Br J Ind Med 38:130– 137.
Berry G, de Klerk NH, Reid A, Ambrosini GL, Fritschi L, Olsen NJ, Merler E, Musk AW. 2004. Malignant pleural and peritoneal mesotheliomas in former miners and millers of crocidolite at Wittenoom, Western Australia. Occup Environ Med 61:e14.
Gardner MJ, Acheson ED, Winter PD. 1982. Mortality from mesothelioma of the pleura during 1968–1978 in England and Wales. Br J Cancer 46:81–88.
Gardner MJ, Jones RD, Pippard EC, Saitoh N. 1985. Mesothelioma of the peritoneum during 1967–1982 in England and Wales. Br J Cancer 51:121–126.
Goldberg M, Imbernon E, Rolland P, Gilg Soit Ilg A, Saves M, de Quillacq A, Frenay C, Chamming’s S, Arveux P, Boutin C, Launoy G, Pairon JC, Astoul P, Galateau-Salle F, Brochard P. 2006. The French National Mesothelioma Surveillance Program. Occup Environ Med 63:390–395.
Greenberg M, Davies TA. 1974. Mesothelioma register 1967–1968. Br J Ind Med 31:91–104.
Harding AH, Darnton A, Wegerdt J, McElvenny D. 2009. Mortality among British asbestos workers undergoing regular medical examina-tions (1971-2005). Occup Environ Med 66:487–495.
Health & Safety Executive. 2009a. Statistics: Data sources: http:// www.hse.gov.uk/statistics/sources.htm.
Health & Safety Executive. 2009b. Statistics: Death certificates mentioning mesothelioma: http://www.hse.gov.uk/statistics/tables/ meso01.htm.
Health & Safety Executive. 2009c. Asbestosis. http://www.hse. gov.uk/ statistics/causdis/asbestosis/index.htm.
Hein MJ, Stayner L, Lehman E, Dement JM. 2007. Follow-up study of chrysotile textile workers: Cohort mortality and exposure-response. Occup Environ Med 64:616–625.
Hutchings S, Jones J, Hodgson J. 1995. Asbestos related diseases. In: Drever F, editor. Occupational health decennial supplement. The Registrar General’s decennial supplement for England and Wales. London: HMSO, p 127–152.
Ilg AG, Bignon J, Valleron AJ. 1998. Estimation of the past and future burden of mortality from mesothelioma in France. Occup Environ Med 55:760–765.
Kjellstrom T, Smartt P. 2000. Increased mesothelioma incidence in New Zealand: The asbestos-cancer epidemic has started. N Z Med J 113: 485–490.
Leigh J, Driscoll T. 2003. Malignant mesothelioma in Australia, 1945– 2002. Int J Occup Environ Health 9:206–217.
Marrett LD, Ellison LF, Dryer D. 2008. Canadian cancer statistics at a glance: Mesothelioma. CMAJ 178:677–678.
Marsh G, Youk A, Sefcik S, Alcorn C. 2004. Occupational Cohort Mortality Analysis Program 4.00 (release 01e) ed. Pittsburgh: Department of Biostatistics, University of Pittsburgh. McDonald JC, Harris JM, Berry G. 2006. Sixty years on: The price of assembling military gas masks in 1940. Occup Environ Med 63:852– 855.
McElvenny DM, Darnton AJ, Price MJ, Hodgson JT. 2005. Meso-thelioma mortality in Great Britain from 1968 to 2001. Occup Med (Lond) 55:79–87.
Mizell KN, Morris CG, Carter JE. 2009. Antemortem diagnosis of asbestosis by screening chest radiograph correlated with postmortem histologic features of asbestosis: A study of 273 cases. J Occup Med Toxicol 4:14.
National Statistics. 2001. Cancer Statistics: Registrations. Registrations of cancer diagnosed in 1995–1997, England. Series MB1 no. 28: The Stationary Office, London. p 96.
Peto J, Rake C, Gilham C, Hatch J. 2009. Occupational, domestic and environmental mesothelioma risks in Britain: Health & Safety Executive. p 63.
Pira E, Pelucchi C, Buffoni L, Palmas A, Turbiglio M, Negri E, Piolatto PG, La Vecchia C. 2005. Cancer mortality in a cohort of asbestos textile workers. Br J Cancer 92:580–586.
Ross D, McDonald JC. 1995. Occupational and geographical factors in the epidemiology of malignant mesothelioma. Monaldi Arch Chest Dis 50:459–463.
Smartt P. 2004. Mortality, morbidity, and asbestosis in New Zealand: The hidden legacy of asbestos exposure. N Z Med J 117:U1153. Stata Corp. 2007. Stata statistical software: Release 10. College Station, TX: StataCorp LP.
