Continuing increase in mesothelioma mortality in Britain

Summary Introduction
Mesothelioma is closely related to exposure to asbestos, Mesothelioma is almost always fatal; most patients
and mesothelioma mortality can be taken as an index of affected die within a year of diagnosis. The majority of
cases are pleural in origin, although peritoneal tumours
past exposure to asbestos in the population. We analysed
are equally common in some groups of workers exposed
mesothelioma mortality since 1968 to assess the current
to amosite (brown) asbestos. The risk is negligible up to
state of the mesothelioma epidemic, and to predict its
future course. 10 years after first asbestos exposure and very low up to
15 years, but increases thereafter as the third or higher
We found that rates of mesothelioma in men formed a 1

clear pattern defined by age and date of birth. Rates rose power of time since first exposure.’
The association of mesothelioma with asbestos was first
steeply with age showing a very similar pattern in all five- noted in 1960,2 and since 1968, the UK Health and
year birth cohorts. By date of birth, rates increased from
mid-1893 to mid-1948, and then fell. Relative to the
Safety Executive has maintained a register of deaths in
1943-48 cohort, the risk for the 1948-53 cohort is 0·79
England, Wales, and Scotland for which mesothelioma is
mentioned on the death certificate.33 Annual deaths
and for the 1953-58 cohort 0·48. Despite these falls, if the increased from 154 in 1968 to 1009 in 1991, and
age profile of rates for these cohorts follows the pattern of increases have also occurred in other countries.4 The
past cohorts, their predicted lifetime mesothelioma risks increase in rates in individuals above the age of 60 is the
will be 1·3%, 1·0%, and 0·6%. Combining projections for all expected continuation of trends observed over the past 20
cohorts results in a peak of annual male mesothelioma years, reflecting asbestos exposure when those now aged
deaths in about the year 2020 of between 2700 and 3300 over 60 began their working lives in the early 1950s and
deaths. If diagnostic trend is responsible for a 20% growth earlier. However, there is a continuing increase in the
in recorded cases every 5 years—an extreme but arguable death rate among men now aged under 50, most of whom
case—and if this trend has now ceased, the peak of annual began work in the mid 1960s or later. The increase
male deaths will be reduced to 1300, reached around the suggests that asbestos exposure was greater around 1970
year 2010. Analysis of occupations recorded on death than in any previous period, and that mesothelioma rates
certificates indicate that building workers, especially will continue to increase as this generation ages. We
plumbers and gas fitters, carpenters and electricians are describe projections of future rates calculated from these
the largest high-risk group. data, and discuss the possibility that occupational
These data indicate that mesothelioma deaths will exposure may still be common, particularly among
continue to increase for at least 15 and more likely 25 building workers.
years. For the worst affected cohorts—men born in the
1940s—mesothelioma may account for around 1% of all Data and methods
Mesothelioma death rates in England, Wales, and Scotland for
deaths. Asbestos exposure at work in construction and
both sexes since 1968 (table 1) and male rates for men born in
building maintenance will account for a large proportion of successive 5-year periods since 1 July 1893 were calculated from
these deaths, and it is important that such workers should the mesothelioma register and population numbers for each year
be aware of the risks and take appropriate precautions. by single years of age.
The distribution of male deaths by age and year of birth (table
Lancet 1995; 345: 535-39
2) was analysed by Poisson regression, fitting the simple age and
birth cohort model: annual age-specific death rate=kacb, where
ka are predicted age-specific rates with a=1 for 25-29, 2 for
30-34, ... 13 for 85-89 and Cb are birth cohort specific relative
risks with b=l for 1893-98, 2 for 1898-1903, ... 13 for
1953-58. The relative risk (RR) for the 1943-48 birth cohort
(c11) was arbitrarily set at unity. The birth cohort RRs are thus
measured relative to the 1943-48 cohort, and the kas are the
predicted age-specific death rates for this cohort. Because the
observation period is defined by calendar year of death, when the
data are presented by year of birth and age at death, as in table 2,
the observations for the youngest and oldest deaths for each birth
cohort are incomplete. Thus for men born between mid-1913
and mid-1918 the youngest deaths during 1968-91 fall in the age
Section of Epidemiology, Institute of Cancer Research,
group 45-49, but all these deaths are at ages 49-5 and more. The
15 Cotswold Road, Belmont, Surrey SM2 5NG, UK (Prof J Peto MSc, men contributing to this cell of table 2 are thus on average nearly
F E Matthews MSc), and Health and Safety Executive, two and a half years older than those contributing to the 45-49
Epidemiology & Medical Statistics Unit, Bootle, Merseyside, UK age group for the 1923-28 cohort, for whom the observations in
(J T Hodgson MSc, J R Jones MSc) this age range are complete. The second youngest and the oldest
Correspondence to: Prof Julian Peto cells for each birth cohort are also incomplete, and have mean

535
Table 1: Numbers of mesotheliomas mentioned on death certificates (Obs) and death rates per million (Rate) in England, Wales,
and Scotland by age and sex 1968-91

ages about 0-7 years higher and 1-3 years lower, respectively, cohort by standard actuarial methods, assuming 1992 British
than the mid-point of the age-group ranges. In the regression male death-rates for all other causes of death in the future
analysis, the fit to cells with incomplete observations was (figures 1 and 2).
iteratively adjusted by log-linear interpolation to allow for their Indirectly age-standardised occupational proportional
difference in mean age from the corresponding complete cells. mortality ratios (PMRs) for male mesothelioma deaths are shown
The estimates ka and Cb were multiplied to give predicted death- in table 3. These were calculated in the standard way according
rates for each age group and birth-cohort. The fitted numbers to the last full-time occupation as recorded on male death
shown in table 2 are the product of these predicted rates and the certificates for ages 16 to 74 during 1979-90 (excluding 1981,
man-years in each cell. The predicted rates were then used to when a strike by registrars made the occupational data less
calculate future mesothelioma deaths in men and the lifetime (to reliable). Complete data and a description of the occupational
age 90) probability of dying of mesothelioma for each birth classification will be published separately.5

Each cell shows the observed number (Italiclsed) of mesothelioma deaths, and the fitted number derived from the birth cohort model by multiplying the population man-years (not
shown) by the product of cb and ka (see methods). *Fltted (to age 49) and predicted (from age 50) death rates per million for 1943-1948 birth cohort. (k,). ftvhd-yearto mid-year.
Table 2: Mesothelioma deaths in British men 1968-91, and fitted numbers based on birth cohort analysis

536
 Results
Age and birth cohort
Close agreement between numbers of mesothelioma
deaths and fitted numbers derived from the age and birth
cohort analysis (table 2) shows that this model accounts
well for the observed trends. Figure 1 shows the
corresponding predicted lifetime probability of dying of
mesothelioma for men born in different periods. The
estimated lifetime risk (to age 90) increases steadily from
0-03% for men born between 1893 and 1898 to 1-3%, or
about 1 in 80, for men born in 1943-48 (mid-year
ranges). The estimated lifetime risks for the last two birth
cohorts are 1.0% (1948-53, based on 39 deaths) and
0-6% (1953-58, based on 7 deaths), but with wide
confidence intervals. Predicted numbers of mesothelioma
deaths in men over the next 50 years based on these
estimates are shown in figure 2. Ignoring the possibility of
secular trends in under-diagnosis or over-diagnosis, these
-chrysotile 0 Amoslte Crocidolite
predictions are reasonably reliable up to about 2020,
when over 70% of all mesothelioma deaths will still be Figure 2: Predicted mesothelioma deaths in British men and
UK asbestos imports
occurring in men born before 1948. After 2020, however,
the prediction will rapidly be dominated by men born One aspect of our analysis suggests some diagnostic
after 1958 for whom no data are yet available. If their risk trend. The mesothelioma rate among USA insulation
is negligible, the epidemic will peak at about 2700 deaths workers’ was approximately proportional to time since
per year and will disappear rapidly after 2020. If their first exposure raised to the power 3-2, and a similar value
lifetime risk is 50% of the 1943-48 maximum, the annual
(3-5) was estimated for Australian crocidolite (blue)
total will peak at about 3300 deaths around 2020 and asbestos minersbut the increase with age of our
then fall to about 2300 deaths per year. The eventual risk
predicted death rates ka(table 2) suggests a power greater
is likely to lie somewhere between these limits. than 4. Assuming an average age at first exposure of 25,
for example, the rates for ages 45-79 estimated from our
Diagnostic trend
The main reservation concerning these predictions is the simple age-cohort model are roughly proportional to (age
minus 25) to the power of 4-5. Refitting the age-cohort
possibility that part of the increase in recorded death rates model with the addition of a diagnostic factor of 1-2 (ie, a
is an artifact of misdiagnosis. If the true rate is kacb for
20% increase in recorded rates from each five-year period
age-group, a, and birth cohort, b, an increase in recorded
to the next) produces a fit to past observations identical to
rates at all ages by a constant factor r from each 5-year
that shown in table 2, but with age coefficients
period to the next due to a continuing increase in
proportional to (age minus 25) to the power of 3-2. A
completeness of diagnosis (or of over-diagnosis) will
mimic an epidemic with age-specific rates proportional to diagnostic factor of around 1-2 thus brings the pattern of
age coefficients more closely in line with the increases in
raka and birth cohort relative risks, rbcb suggesting a
spuriously steep increase in risk, both with increasing age
and in successive birth cohorts. Such a secular trend in
diagnosis would be statistically indistinguishable from a
real increase, but would cease as soon as reasonably
complete diagnosis was achieved.
c 1.R -

*p<0001, tp<0.01, fp<0.05. Highest 25 occupational PMRs based on 10 or more

Year ot oirtn (mia-year to mia-year) deaths

Figure 1: Predicted lifetime probability of dying from Table 3: Proportional mortality ratios (PMR) of men aged
mesothelioma for British men by year of birth (95% confidence 16-74 from mesothelioma in England and Wales 1979-80,
intervals) 1982-90

537
risk with time since first exposure seen in other cohort from men whose asbestos exposure often continued
studies. throughout their working lives. If asbestos exposure fell
Projections with these lower age coefficients, together sharply after (say) 1980, this will not be the case for men
with the assumption that no further diagnostic increases born after 1940, because their exposure will have largely
will occur after 1995, would imply that the number of ceased by the age of 40. The rates for this and subsequent
male mesothelioma deaths will rise much more slowly, cohorts may therefore not continue to rise as steeply as in
reaching a peak of about 1300 around the year 2010, and earlier generations. There is as yet no direct evidence of
about 1500 per year including women. any change in the pattern of age dependence, but the
lower rates recorded for the 1948-53 and 1953-58 birth
Occupations at risk cohorts indicate that population exposure has indeed
The proportional mortality ratios (PMR) for different fallen.
occupations in table 3 show the highest risk in metal-plate Irrespective of future trends and past errors of
workers (the occupational category which includes diagnosis, total asbestos-related cancer deaths in the UK
shipyard workers) and vehicle body builders. These two must be substantially greater than the number of recorded
occupational categories accounted for 3% of all male mesothelioma deaths. As noted above, diagnostic review
mesothelioma deaths. The next three highest PMRs are of deaths among asbestos workers increases the number of
for plumbers and gas fitters, carpenters, and electricians, deaths attributed to mesothelioma, especially those
and a further six of the 25 occupational groups shown in originating in the peritoneum. Cohort studies of workers
table 3 are also in construction or related trades. These heavily exposed to asbestos also show an excess of lung
nine groups account for 1083 deaths, or 24% of the total. cancer similar to the mesothelioma risk for crocidolite and

greater for other types of asbestos.4

Discussion Since the carcinogenic effects of asbestos were widely
The prediction that British mesothelioma rates will recognised by the mid 1960s, it is surprising that the
continue to rise for many years is fairly secure. Whether worst effects should have been experienced by the

the eventual total will be of the order of 3000 deaths per generation of men who began work after this date. The
year, as the simple cohort analysis suggests, depends on explanation is that most exposures (not the most intense,
several factors. Our analysis is based on death certificates but affecting large numbers) occurred in occupational
on which mesothelioma is mentioned, with no systematic settings, particularly in the building industry, which were
histological review. Misdiagnosis was certainly common in and still are largely unmonitored. After introduction in
the past, particularly for peritoneal mesotheliomas; the UK of the 1969 Asbestos Regulations, HM Factory
pathological review of deaths occurring among insulation Inspectorate (HMFI) adopted the hygiene standard
workers in the USA up to 1976 increased the number of recommended by the British Occupational Hygiene
peritoneal mesotheliomas from 23 recorded on the Society (1968),10 which was 2 fibres/mL for chrysotile
original death certificate to 112.7 The corresponding (white), amosite, and fibrous anthophyllite asbestos. On
figures for pleural mesothelioma were 37 by death the basis of its link with mesothelioma, a more stringent
certification and 63 after review. The UK misdiagnosis standard of 0-2 fibres/mL was adopted for crocidolite.11 In
rate during the 1970s was also substantial. 38% (17/45) 1971, the Medical Services Division of HMFI established
of mesothelioma deaths among workers in a London a prospective mortality study of men in a limited number

asbestos factory were originally attributed to other of workplaces which were covered by the 1969 Asbestos
cancers.8 There may also have been some degree of over- Regulations, subsequently extended to cover most fixed
diagnosis in recent years.9 There are, however, several workplaces, and in 1986 to all individuals having statutory
qualitative arguments which suggest that diagnostic trends medical examinations under the Asbestos Licensing
cannot account for a large part of the recent increase in Regulations.12 183 mesothelioma deaths occurred in this
recorded rates. First, the increase in rates from 1982-86 cohort from 1971 to 91. Over the same period, 10 985
to 1987-91 was as great as in earlier periods. Such a mesothelioma deaths occurred nationally; this figure
striking diagnostic increase over the last decade seems suggests that the vast majority of workers actually at risk
unlikely. Second, mesothelioma rates in cohort studies of from asbestos were not employed in occupations where
asbestos workers continue to rise with increasing time this risk was recognised. 24% of male mesothelioma
since first exposure. In contrast, the national mortality deaths are listed under construction-related occupations
data suggest a roughly constant death-rate above age 55 in table 3, but the proportion due to exposure in the
for 1968-71 and now show a peak in the death-rate at construction industry is probably considerably higher.
about age 75 (table 1), a pattern characteristic of cross- Table 3 is based on the most recent full-time occupation
sectional rates in a progressive epidemic. The strongest as recorded on the death certificate, so men who leave the

reason for accepting that the generation of British men building industry would not be included in this number.
born between 1940 and 1950 may have a high risk of Asbestos imports were at their peak in the 1960s and
mesothelioma is that the first 10-20 years of their working 1970s (figure 2). In that period the major use of
lives coincided with the peak of asbestos imports to amphibole asbestos was in amosite insulation board used
Britain during the 1960s and 1970s (figure 2). But in the construction industry, where monitoring and
although improvements in diagnosis cannot entirely control of exposure were limited. In addition, more than
account for the observed increase in rates, even a small half the tonnage of chrysotile imported went into
diagnostic trend could substantially affect the long-term products for the construction industry, mainly asbestos
projections. Our long-term predictions also depend on the cement products and floor tiles.
assumption that the death-rate in men born since about Most of the asbestos imported to the UK between 1960
1940 will continue to increase with increasing age as and 1980 is still in place in buildings, and carpenters,
sharply as in earlier generations. The relationship between plumbers, electricians and other workers involved in
age and the predicted death rate shown in table 2 arises building renovation, maintenance, and demolition may

538
still suffer unsuspected exposure. These workers often The risk to recent birth cohorts could also be modified
operate -individually or in small unmonitored by changes in the prevalence of infection with the simian
organisations, and it may be impossible to obtain reliable virus SV40, which was a contaminant of poliovaccines
information on the extent of their exposure. A public which were widely used in the 1950s. According to a
information campaign to alert such workers to the recent report, 15 SV40-like DNA are present in
sequences
potential danger seems justified. most pleural mesotheliomas. Most mesotheliomas
We have confined our analysis to male rates, because containing SV40-like DNA were also associated with
there are too few deaths in women at younger ages to asbestos exposure, so this effect, if confirmed, is likely to
provide stable predictions of future mortality. Trends in be synergistic with that of asbestos.
female rates over the last 25 years have been similar to The Institute of Cancer Research receives financial support from the
those in men, although at a much lower level, as would be Cancer Research Campaign. We acknowledge the contribution of support
staff in HSE and OPCS for assembling and managing this data over many
expected for an occupational cancer (table 1). Death-rates years.
in UK women in 1968-71 were similar to incidence rates
in Los Angeles in the 1970s in both men and women with
References
no suspected asbestos exposure. 13 This may reflect a
1 Peto J, Seidman H, Selikoff IJ. Mesothelioma mortality in asbestos
spontaneous background risk unrelated to asbestos, workers: implications for models of carcinogenesis and risk assessment.
although some cases in both sexes may be due to Br J Cancer 1982; 45: 124-35.
environmental asbestos exposure unrelated to 2 Wagner JC, Sleggs CA, Marchand P. Diffuse pleural mesothelioma and
asbestos exposure in the North Western Cape Province. Br J Indust
occupation. Med 1960; 17: 266-71.
The eventual magnitude of the British mesothelioma 3 Jones RD, Smith D, Thomas PG. Mesothelioma in Great Britain
epidemic is likely to be greater than in the USA. There is 1968-83. Scand J Work Environ Health 1988; 14: 145-52.
no national mesothelioma death registry in the USA, but 4 Health Effects Institute&mdash;Asbestos Research. Asbestos in public and
commercial buildings: a literature review and synthesis of current
the SEER cancer incidence data indicate that the USA
knowledge. Cambridge, MA: Health Effects Institute, 1991.
epidemic has already reached its peak.4 Overall incidence 5 Office of Population Censuses and Surveys/Health and Safety
increased rapidly during the 1970’s, and in the early Executive. Occupational Health Decennial Supplement. London:
HM Stationery Office, 1995 (in press).
1980s male incidence rates were similar to UK death rates
6 Berry G. Prediction of mesothelioma, lung cancer, and asbestosis in
in men aged over 65. Since 1980, USA rates have former Wittenoom asbestos workers. Br J Ind Med 1991; 48: 793-802.
declined in both sexes below age 55, however, and are no 7 Selikoff IJ, Hammond EC, Seidman H. Mortality experience of
longer increasing in men aged under 75, in contrast to the insulation workers in the US and Canada, 1943-76. Ann NY Acad Sci
1979; 330: 91-116.
pattern in the UK. Overall numbers in the USA are 8 Newhouse ML, Berry G, Wagner JC. Mortality of factory workers in
therefore likely to fall over the next 20 years. The timing east London 1933-80. Br J Ind Med 1985; 42: 4-11.
of the peak in the USA epidemic reflects the pattern of 9 Dini S, Santucci M, Biancalani M, et al. Pleural malignant
asbestos use, which reached a plateau soon after World mesothelioma in Tuscany, Italy (1970-1988): I. Anatomo-pathologic
aspects. Am J Ind Med 1992; 21: 569-76.
War II.13 10 British Occupational Hygiene Society. Hygiene standards for chrysotile
Any contribution of asbestos removal to the asbestos dust. Ann Occup Hygiene 1968; 11: 47-69.
mesothelioma epidemic cannot yet be assessed. Asbestos 11 HMFI Technical Data Note 13, Department of Employment and
removal did not develop as a specialised industry until the Productivity. Standards for asbestos dust concentration for use with the
Asbestos Regulations 1969. Her Majesty’s Factory Inspectorate.
1980s, and the latency is still too short for these workers London: HM Stationery Office; 1970.
to have developed mesothelioma. The campaign to 12 Hodgson JT, Jones RD. Mortality of asbestos workers in England and
remove all asbestos from schools and other public Wales 1971-81. Br J Ind Med 1986; 43: 158-64.
13 Peto J, Henderson BE, Pike MC. Trends in mesothelioma incidence in
buildings on the grounds that occupants might be at high the United States and the forecast epidemic due to asbestos exposure
risk was founded on fear rather than evidence. Average during World War II. In: Peto R, Schneiderman M, eds. Quantification
airborne asbestos levels in such buildings are invariably of occupational cancer. Banbury Report 9, Cold Spring Harbor
Laboratory, 1981.
very low during normal use, and are often unaffected by 14 Burdett GJ, Jaffrey SAMT, Rood AP. Airborne asbestos fibre levels in
removal and may even be increased.’4 Whatever its effect buildings&mdash;a summary of UK measurements. In: Bignon J, Peto J,
on building occupants however, the creation of the new, Saracci R, eds. Non-occupational exposure to mineral fibres. IARC
Scientific Publications no 90. Lyon: International Agency for Research
and initially inadequately regulated, industry of asbestos
on Cancer, 1989: 277-90.
removal may well have increased the burden of future 15 Carbone M, Pass HI, Rizzo P, et al. Simian virus 40-like DNA
occupational asbestos disease. sequences in human pleural mesothelioma. Oncogene 1994; 9: 781-90.

539