INT J TUBERC LUNG DIS 11(7):722–732
© 2007 The Union
Historical statistics support a hypothesis linking
tuberculosis and air pollution caused by coal
G. A. Tremblay
Phoresia Biotechnology Inc., Laval, Québec, Canada
Tuberculosis (TB) is generally considered to be linked to
industrialisation and urbanisation. Peaking in the 1800s
and receding slowly after, the disease declined sharply in
the West after World War II. TB has made a comeback in
the last 20 years in developing countries such as China
and India. Because socio-economic conditions alone can-
not explain the connection between industrialisation and
TB, factors remain to be determined in the aetiology of
the disease. Historical statistics on coal consumption and
THE INVENTION of the blast furnace and the steam
engine were key developments that led to industriali-
sation. From that point on, fossil fuel has driven the
evolution of the world as we know it today; with
progress and technical innovations, changes in life-
style ensued promptly. However, as humans unlocked
their creative potential in mechanics, they may also
have unleashed the avatar of progress.
A century-old belief blamed foul air or miasmas
for diseases such as tuberculosis (TB), malaria, rheu-
matism and cholera. A number of scientists, including
Florence Nightingale and William Farr, supported the
theory. Nevertheless, miasmas fell into disgrace when
Louis Pasteur discarded spontaneous generation and
it became clear that micro-organisms were largely re-
sponsible for mankind’s ills.
Mycobacterium tuberculosis, the agent responsible
for TB—also called phthisis or consumption—was
identified in 1882 by the German physician Robert
Koch. At the time, TB claimed about one death in six
in Europe.
TB has made a comeback in the last 20 years, mostly
in parts of the developing world. In 2004, 9 million
new TB cases were diagnosed and approximately 2
million persons died of TB around the world. More
than 80% of all TB patients live in sub-Saharan Africa
and Asia. People infected with the human immunode-
ficiency virus (HIV) are more susceptible to TB.1 In Af-
rica, for example, HIV is said to drive the TB epidemic.2
TB has been termed a ‘social disease’ because it is
linked to poverty, overcrowding and unsanitary con-
Correspondence to: Guy A Tremblay, 3126, 5ième Rue, Laval, Québec H7V 1M1, Canada. Tel: (1) 514 436 9272. e-mail:
guy.tremblay@gmail.com
Article submitted 24 September 2006. Final version accepted 26 January 2007.
REVIEW ARTICLE
SUMMARY
TB disease in Canada, USA and China are correlated.
A hypothesis linking TB and air pollution is developed
in the context of industrialisation. A model is proposed
whereby triggering of the interleukin-10 (IL-10) cascade
by carbon monoxide in lung macrophages promotes the
reactivation of Mycobacterium tuberculosis.
K E Y W O R D S : tuberculosis; coal; air pollution; carbon
monoxide; IL-10
ditions. Improvements in living conditions are there-
fore likely to affect TB incidence.
There is general agreement that the TB epidemic
increased dramatically in Europe in the eighteenth
and nineteenth centuries and began to decline there-
after.3 TB is intrinsically linked to industrialisation.
For example, both industrialisation and TB occurred
later in Japan.4 As there is no obvious reason to be-
lieve that there was less poverty before industrialisa-
tion, and considering that overcrowding is relevant to
all communicable diseases, one is forced to question
why industrialisation and TB go together.
Another unresolved issue concerns the decline of
TB in the West. For example, between 1850 and 1910,
TB death rates declined in England and Wales much
faster than indicators of social deprivation.5 It seems
therefore that social factors alone cannot explain the
decline in TB incidence.
There may in fact have been some degree of confu-
sion between social and environmental factors. For
example, in a 1911 study,6 Marie-Davy compared TB
mortality and the average number of windows per
household (Figure 1) in Paris arrondissements (urban
districts) between 1858 and 1902. Scientists of the
time were aware that inadequate aeration of houses
was related to the development of TB in residents;
Figure 1 may also indicate high levels of indoor air
pollution in urban households.
In recent years, air pollution has been identified as
a risk factor for TB. Mishra et al. linked biomass
cooking fuels and TB:7 based on a 1992–1993 survey

Figure 1 TB mortality rate as related to the average number of
windows per household in Paris urban districts or arrondisse-
ments between 1858 and 1902. pop  population.
on 90 000 households, the study provides evidence
that women in biomass-using households are three
times more likely to report TB than those in households
using cleaner fuels. Another Indian study found that
households using wood or dung cakes as fuel had 2.5
times more clinically confirmed TB.8 Cigarette smoking
is now recognised as a risk factor for TB infection.9
Miners in coal and gold mines are also prone to TB, al-
though it is considered an effect of silicosis.10,11
Based on the observation that industrialisation and
TB coincide, and on the assumption that a determi-
nant factor remains to be identified in the TB epi-
demic, historical statistics of disease and energy were
compared to assess the impact of environmental fac-
tors on TB.
The present hypothesis suggests that pollution
caused by combustion of coal and possibly town gas
during industrialisation aggravated the historical TB
epidemic in the West and the present epidemic in de-
veloping parts of the world. The possibility that car-
bon monoxide (CO) and particulate matter (PM) ac-
tivate the interleukin-10 (IL-10) anti-inflammatory
Table Pearson partial correlation coefficients relating disease incidence or notification rate
(China), electrification and indicators of coal consumption in Canada, USA and China
Canada
Employees in
Disease incidence the coal
per 100 000 industry
population (1940–1975)
TB 0.97
Typhoid fever 0.87
Whooping cough 0.85
Diphtheria 0.77
Scarlet fever 0.55
Poliomyelitis‡ 0.31
* Correlation in the US for residential coal is done with TB incidence/100 000 population. When using TB death rate/
100 000 in the US, the correlation was also 0.99.
† In China, TB notification rate/100 000 is used.
‡ Correlations for the period 1940–1964.
TB  tuberculosis.
Air pollution caused by coal impacts on tuberculosis 723
cascade is discussed. The observations shed new light
on TB biology and epidemiology and emphasise the
importance of environmental health.
METHODS
Historical statistics of disease in Canada, as incidence
per 100 000 population, are gathered from Statistics
Canada Historical Statistics on ‘Annual rates of notifi-
able diseases, Canada, 1926 to 1975’ (series B517–
525).12 For electricity production, ‘Electric utilities—
number of customers by class, 1920 to 1975’ (series
Q97–101) is used.12 Electricity production is ex-
pressed as percentage of maximum customers over the
period interval. ‘Principal statistics of the Canadian coal
mining industry, 1918 to 1976’ (series Q137–142) is
used for the numbers of employees in the coal indus-
try.12 Number of coal employees is expressed as per-
centage of maximum over the period. Bacille Calmette-
Guérin (BCG) vaccination in Canada, expressed as total
number of vaccine shots, was compiled by Malissard.13
US Department of Energy Official Energy Statistics
provides historical statistics on coal consumption by
sector in million short tons from 1949 to 2005,14 and
the Centers for Disease Control and Prevention (CDC)
Division on Tuberculosis Elimination Statistics data
on historical TB in the US (incidence/100 000 between
1953 and 2005).15
Data on coal consumption (exajoules) in China are
taken from the China Statistical Yearbook 2003,16
and TB notification rate (/100 000) from the WHO TB
report 2005.17 The poverty headcount index repre-
sents the percentage of the population living in house-
holds with income per person below the poverty line.
The headcount index is provided by the World Bank
Policy Research Working Paper Series.18
Pearson correlation coefficients, R2 and P number
were calculated using Microsoft Excel mathematical
functions. Pearson correlation is carried out with data
from the corresponding figure for the given time period
USA China
Electricity, Residential coal Total coal
total number consumption, consumption
of customers million short tons in exajoules
(1940–1975) (1953–1990) (1982–2002)
0.95 0.99* 0.95†
0.86
0.87
0.76
0.55
0.21
724 The International Journal of Tuberculosis and Lung Disease

in the Table. Steel industry numbers, population data RESULTS

and electric drive figures for Norway are from Statis-
tike Centralbyrå,19,20 whereas TB mortality figures
(/100 000) are taken from Wood.21 The progress of
urban gas lighting (total number of gas lights installed)
in Paris is from Bouteville,22 whereas TB mortality
figures (/1000) in Paris are derived from Barnes.23
Waldron provides the number of notified workplace
accidents during World War II,24 and TB historical
statistics (total number of cases) for England and
Wales are from the United Kingdom’s Statutory Noti-
fications of Infectious Diseases (NOID).25
TB decline in twentieth century North America
Figure 2 shows the incidence of notified diseases and
BCG vaccination in Canada between 1926 and 1975.
There was a global downward trend of all diseases over
the period. TB incidence rose after 1930 (Figure 2A),
increased further with the advent of WWII to peak in
1944, and declined steadily after the war. In contrast,
whooping cough increased around 1935, but not at
the onset of the war (Figure 2C). Scarlet fever shows
cyclic behaviour every 8 years or so (Figure 2D). An

Figure 2 Incidence of notified disease (A, C–F) in Canada during the twentieth century. Vertical dotted
lines highlight WW II. In B, BCG vaccination in the province of Quebec (solid line), in the province of New-
foundland (dotted line) and in the rest of Canada (dashed line) are shown. pop  population.
 Air pollution caused by coal impacts on tuberculosis 725

outbreak of typhoid fever occurred in 1927; it de-

clined sharply then steadily afterwards, to almost dis-
appear (Figure 2E). Diphtheria also decreased after
1930, came back to some extent before WWII and
virtually disappeared after the war, presumably as a
consequence of vaccination (Figure 2F).
Of all the diseases shown in Figure 2, TB is perhaps
the disease most closely associated with WWII. This
can be explained by the arrival of the mass workforce
in manufactures involved in the war effort, considering
that factory and construction labourers were the prime
targets of TB in the 1930s.26 The cyclic behaviour of
scarlet fever was somewhat altered by the war, and
the increase in diphtheria may also have something to
do with living and working conditions during the war.
The number of BCG vaccine shots administered in
the province of Quebec, Newfoundland and the rest of
Canada are shown in Figure 2B. It is likely that BCG
vaccination had only a limited impact on the decline of
pulmonary TB in Canada and elsewhere after WWII.
First, TB decreased in countries where BCG was not ad-
ministered, such as the USA and the Netherlands. Sec-
ond, in Canada, the vaccine was mostly administered in
Quebec and Newfoundland (Figure 2B), and only ex-
ceptionally in the rest of Canada, whereas no bound-
aries were observed in TB incidence. Finally, BCG is a Figure 3 A. TB incidence (solid line), total number of coal
employees in the coal industry (dashed line) and electric utility
controversial vaccine, with an efficacy that has been re- customers (dotted line) in Canada between 1940 and 1975.
ported to vary between 0 and 80%.27 In addition, three B. Historical US domestic coal consumption (dashed line) and TB
meta-analyses carried out in the mid-1990s have shown incidence (solid line) between 1953 and 2003. pop  population.
a protective overall effect of 50%,28–30 with greater ef-
ficacy for disseminated forms and meningitis but lesser
efficacy in the case of pulmonary TB. was not administered (Figure 3B). Data for residential
As regards anti-tuberculosis drugs, streptomycin coal consumption in the USA are available. TB de-
(SM) did not become available before 1948, and resis- clines in the US again coincide with a reduction in res-
tance soon developed. Introduced in 1952, isoniazid idential coal consumption. A peak in TB incidence in
reduced resistance when used in combination with 1977 is visible and the increase in TB due to the HIV
SM.31 However, the decline in TB incidence in 1952 is epidemic between 1990 and 1997 can be seen. As for
not sharper than before, as could have been expected residential coal, its use has almost completely disap-
if the anti-tuberculosis drugs had had 100% effect peared by 1980.
(Figure 2A), and if it was administered systematically.
Figure 3A shows the number of coal workers, the Modern-day TB epidemic in China
number of customers of electric utilities and TB rates Overall coal consumption in China,16 when com-
in Canada after WWII, based on data obtained from pared to TB notification rates,17 shows an apparent
Statistics Canada.32 One important advance of the post- close relationship between global increases in coal con-
war era was the electrification of homes. In Canada, sumption and TB disease notification rates (Figure 4A).
gasoline and electricity replaced coal as the principle TB incidence and coal use have increased in a similar
means of heating households. The relative number of fashion in the last 20 years, with simultaneous peaks
coal workers during that period is used to illustrate around 1986, 1990 and 1996, and a corresponding
the decline of the coal industry in Canada, mainly due drop around 2000. In sharp contrast to trends in TB
to the loss of the residential market. In Canada, TB notifications and coal consumption in China, Figure
declined to levels below 20/100 000. This drop pre- 4B shows that overall poverty for rural and urban
cedes that of coal employees, which may reflect a de- areas decreased during the same period in China.18
lay in the lay off of the workforce relative to the evo- Social deprivation cannot, therefore, explain the in-
lution of the coal market. Nonetheless, the timelines crease in TB notifications (Figure 4A).
of TB decline and the indicators of residential coal China is the largest consumer and producer of coal in
consumption do coincide. the world. Today, coal is widely used in China as a
A similar pattern of TB decline is observed be- source of energy for both electricity production and
tween 1953 and 1975 in the United States, where BCG household use.33 Coal largely dominates the energy
726 The International Journal of Tuberculosis and Lung Disease
Figure 4 A. Tuberculosis notification rates (solid line) and total
coal consumption (dashed line) in China between 1980 and 2002.
B. Poverty in China as a whole, in urban and rural areas: percent-
age of the population living in households with income per person
below the poverty line between 1981 and 2001. pop  population.
landscape of the country;16 estimates suggest that 330
million people use coal for domestic purposes.34 Ac-
cording to Finkelman et al., coal stoves and small coal
boilers provide more than 50% of the energy for urban
households, and 22% of rural households rely on coal.33
Low-grade coal is often burned in cheap, poorly vented
or even unvented stoves, resulting in various illnesses.33
Correlation between energy and disease
The Table correlates notified disease, coal employees
and electrification in Canada between 1940 and 1975,
and coal consumption and TB disease indicators in
China and the USA. The Pearson correlations are de-
rived from the data used in Figures 2 and 3.
In Canada, although the onset of TB precedes indi-
cators of residential coal consumption, the Pearson
partial correlation coefficient of 0.97 is strongest com-
pared to other notified diseases for the same period. It
is interesting to note that the negative correlation for
electricity of 0.95 is strongest for TB compared to
other diseases, as is the case for coal. Also, the corre-
lations for electric utility customers are similar in ab-
solute value to those of coal employees for notified dis-
ease in Canada, indirectly indicating that the electricity
market develops at the expense of the coal market.
In the US, again, a very strong correlation of 0.99
is found linking residential coal consumption and TB
incidence between 1953 and 1990. The correlation of
0.95 between total coal consumption and TB notifi-
cation rate in China is also high.
The strong correlations in Canada, USA and China
between coal consumption and TB disease therefore
highlight the possibility that coal consumption directly
impacts on TB disease.
Coal and industrialisation
Before industrialisation in Europe, wood became scarce
and expensive, while coal became affordable. Char-
coal and coal combustion yield more energy than
wood; inventions and innovations led to the blast fur-
nace and the steam engine.
One key feature of industrialisation is the develop-
ment of the steel industry. The advent of this industry
may be considered a consequence of innovations in
the process of iron smelting. The blast furnace is a
metallurgical furnace used for smelting, notably iron
ore. Charcoal or coke, limestone and iron ore are fed
into the blast furnace while air is blasted at the bot-
tom. Iron is reduced to metal by charcoal, which
yields oxides of carbon as products of combustion.
Beginning in the UK in 1709, the use of coke (baked
bituminous coal) instead of charcoal (baked wood) in
blast furnaces was of paramount importance in the
advent of the industrial revolution, as it enabled the pro-
duction of iron that was cheaper and less brittle. In turn,
the availability of coal on the market spurred the use of
and technological innovations on the steam engine, as
well as the development of heavy industry. This serves to
illustrate how industrialisation boils down to coal.
TB and industrialisation
There is some agreement that TB incidence peaked
around 1800 in Western Europe.35 TB accounted for
25% of deaths in New York city between 1810 and
1815.36 However, according to Barnes,23 TB in Paris
peaked around 1865, whereas in Norway it peaked
around 1900.21 Therefore, based on the present hy-
pothesis, the peak in TB disease depends on the extent
of air pollution and industrialisation in an area; in
China, for example, industrialisation may be consid-
ered to be occurring now.
TB numbers before the 1900s are difficult to find.
In Norway21 and Japan,4 for example, industrialisa-
tion occurred later than in the rest of Europe, as did
the rise in TB incidence.
Figure 5 shows TB incidence between 1855 and 1945
in Norway. Beginning slowly around 1840, Norway in-
dustrialised its pulp and paper industry and introduced
steel works. The number of steel workers jumped from
1608 to 7161 employees and from 9.3% to 22.8% of
the total workforce between 1860 and 1870.19 The
late onset of industrialisation, exemplified by the emer-
gence of the steel industry in Norway (Figure 5), co-
incides with a delayed increase in TB incidence.

Figure 5 Mortality due to TB in Norway between 1855 and
1945 (solid line) with indicators of industrial development
shown on the timeline (diamonds). Electric motor drive in the
manufacturing industry between 1900 and 1938 in Norway
(dashed line). pop  population.
Another idiosyncrasy of Norway in regard to in-
dustrialisation is the early electrification of the country,
early in the twentieth century, due to the abundance
of large water basins; the capacity of water plants
jumped from 10 MW in 1907 to 1250 MW in 1920
and to 2250 MW in 1939.37
The TB epidemic receded earlier in Norway than
in other industrialised countries such as the UK, Can-
ada or the US. As illustrated in Figure 5, early electri-
fication, as exemplified by the use of electric motor
drive in the manufacturing industry between 1900
and 1938, may explain the early decline in TB preva-
lence in Norway, although the decline seemed to start
slightly before the onset of electrification. Neverthe-
less, electrification proceeded very rapidly in Norway.
Again, as was the case in Canada and the US, the
switch to a cleaner energy coincides with a dramatic
decrease in the prevalence of TB.
In the late nineteenth century, Japan opened to the
world following the Meiji restoration, and reforms
ensued. The silk industry financed the Meiji restora-
tion to some extent. During the late nineteenth and early
twentieth centuries, the industrial production of silk
in Japan was revolutionised by the use of steam boilers
for continuous, even heating of the reeling basins used
to soften cocoons to be reeled into raw silk;38 coal was
used to heat the steam boilers.39 TB incidence was high
in the silk and spinning industry, and the young girls
employed for spinning were the first victims. Later, a
second phase of heavy industrialisation in the 1930s
shifted the trend, and young men became the prime
targets of TB (T Shimao, personal communication).
Back in Europe, in the early nineteenth century,
technical advances by the Scottish engineer William
Murdoch and others introduced town gas, also called
synthetic gas or syngas. Town gas was produced by
gasification, a process whereby coal is heated in the
presence of water vapour. Syngas produced with coal
in processes with water-coal slurry systems is typically
Air pollution caused by coal impacts on tuberculosis 727
and mainly composed of (vol%) H2 27–30%, CO 35–
45%, carbon dioxide (CO2) 10–15%, H2O 15–25%,
hydrogen sulphide (H2S) 0.2–1.2% and hydrogen
chloride (HCl) 50–500 ppm.40
A passage in the book by Flora Tristan describes a
scene at the coal gasification factory, the Gas Light
and Coke Company’s Horseferry Road Works, in
London in the 1830s:
Two rows of furnaces on each side were fired up [. . .]
The foreman told me that stokers were selected from
among the strongest, but that nevertheless they all
became consumptive after seven or eight years of toil
and died of pulmonary consumption.41
Town gas was first used for lighting streets and homes
using the gaslight or ‘bec de gaz’. The first gaslights
had a notoriously bad reputation: they were very in-
efficient and leaked large amounts of unburned gas
that added to the coal fumes in urban areas. An ac-
count by Lecouturier depicts industrialised Paris around
the 1850s:
If you contemplate from the summit of Montmartre
the congestion of houses piled up at every point of a
vast horizon, what do you observe? . . . One is tempted
to wonder whether this is Paris; and, seized with a
sudden fear, one is reluctant to venture into this vast
maze, in which a million beings jostle one another,
where the air, vitiated by unhealthy effluvia, rising in
a poisonous cloud, almost obscures the sun.23
Figure 6A depicts the progress of urban gas lighting
in Paris—la ville lumière (the city of light)—during
the nineteenth century.22 Between 1831 and 1852, all
13 000 urban lighting oil lamps of Paris were replaced
by gaslights as gas pipelines were installed for urban,
industrial and domestic usage.
In Paris, the TB epidemic peaked in the middle of
the century, around 1865 (Figure 6B), and receded
slowly thereafter.23 It may be considered that a critical
factor for the acute crisis in Paris is the introduction
of gaslight. As town gas allowed night shifts in manu-
factures, it is likely that TB incidence increased among
factory labourers during the course of the nineteenth
century. In defence of the thesis is the shift in sex ratio
for TB during that period: in 1830, two thirds of
women suffered from TB, while in 1860 the trend re-
versed and two thirds of men suffered from TB;23 men
then constituted the majority of the labour force in
manufactures. The peak of the TB epidemic in Paris
around 1865 can therefore be explained by the wide-
spread use of town gas during that period. Thus, indi-
rectly because of the advent of night shifts in the indus-
try and directly due to town gas fumes, the utilisation
of combustible town gas is likely to have had an im-
pact on TB incidence during the nineteenth century.
The decline in TB mortality towards the end of
that century may be explained by improved combus-
tion and lighting technologies, such as the Auer lamp
invented by Dr Carl Auer von Welsbach in 1885,
728 The International Journal of Tuberculosis and Lung Disease

dents and the number of TB notifications in the UK

Figure 6 A. Total numbers of urban lighting oil lamps (grey)
and town gas lamps (black) installed in Paris in 1831, 1848 and
1852. B. TB mortality rates in Paris during the nineteenth cen-
tury. pop  population.
during WWII, showing a concomitant increase in
workplace accidents and TB in the UK during this
time. Generally speaking, there was an increase in all
types of workplace accidents. How can this be ex-
plained? According to Waldron, work was arduous in
UK factories during the war:
. . . a large and often inexperienced workforce spent
long hours in factories that were often poorly lit,
badly heated and badly maintained. The requirement
for complete blackout meant that ventilation systems
were often inadequate, and this resulted in greater
exposure than normal to some toxic substances. [. . .]
there is no question that the number of cases of spe-
cific industrial diseases increased substantially [. . .].24
It is possible that the increase in TB in the West during
WWII is due to working conditions in manufactures
involved in the war effort. In this regard, TB therefore
translates into an industrial disease, considering the
high rates of TB in construction and factory labourers
in the 1930s. It also highlights the possibility that a
specific workplace pollutant is involved in reactiva-
tion of the disease. Generally speaking, if poverty was
such an important factor in the incidence of TB, the
disease would have increased mostly in the 1930s, as
poverty was rampant during the Great Depression.
Considering that ventilation in the workplace was
less than satisfactory in the first half of the twentieth

whose use spread around 1890. It may be noteworthy

to mention that town gas is still used in industrialising
parts of the world today where natural gas is rare or
expensive, such as in China or South Africa.42
TB, industry and ventilation
In a 1930 occupational hazards study, construction
and factory labourers were found to be the prime tar-
get of TB, with a death rate of 227.3/100 000 com-
pared to 180.0 for waiters, 173.3 for servant classes,
165.6 for teamsters and carriage drivers, 106.0 for
coal mine operatives and 23.7 for physicians and sur-
geons,26 trends that are in sharp contrast with today’s
occupational statistics.
The prevalence of TB in industry and manufac-
tures is a recurrent theme during the nineteenth and
twentieth centuries. A 1914 paper by Hastings directly
links TB and air pollution:
Among garment workers many cases of tuberculosis
develop. This is due largely to long hours, dusty, badly
ventilated, overcrowded rooms, in which they fre-
quently use gas stoves, or gas flat irons and charcoal
stoves. Cases of chronic gas poisoning are not unusual
among these people from leaky gas stoves.43
TB made a comeback in the West during WWII. The
war effort brought masses of workers into manufac- Figure 7 Reported workplace accidents in factories and tuber-
culosis notifications between 1937 and 1946 in the UK, show-
tures. Figure 7A shows carbon monoxide, nitrous ing A) number of carbon monoxide (solid line), nitrous fumes
fumes and lead intoxications in UK manufactures, (dashed line) and lead (dotted line) poisoning notifications; and
while Figures 7A and 7B compare intoxication acci- B) total number of pulmonary TB notifications.

century, it is conceivable that the high incidence of TB
in manufactures is explained by workplace air pollut-
ants. Electricity gradually replaced internal combus-
tion engines as the source of mechanical drive, and
proper ventilation systems were installed in the indus-
try. These changes occurred in parallel to the shift in
household energy sources, thereby reducing the total
burden of air pollution.
Pertaining to the issue of ventilation, ‘open-air treat-
ment’ or ‘sanatorium treatment’ was introduced in
Germany by Hermann Brehmer in Görbersdorf, Sile-
sia, in 1854.44 Although some variations existed be-
tween sanatoria (e.g., rest vs. exercise, altitude vs. sea
level), open-air treatment was essentially based on ex-
posing the patient to fresh air, night and day, in a re-
mote area. In studies of sanatoria in England, it has
been said that patients increased in weight within the
first week (2–6 pounds), cough was seldom heard, and
night sweats tended to cease within days.45,46 Sanatoria
were generally considered beneficial. Strangely enough,
no comparative study on the efficacy of sanatoria was
ever carried out. Bardswell made a careful follow-up
study of patients admitted to the King Edward VII
sanatorium in Sussex between 1907 and 1914.47 He
found that by 1916, 751 of 1707 (44%) patients had
died, making sanatoria less than a panacea. In retro-
spect, and in the light of the present hypothesis, the
sanatorium treatment may be regarded as a way to
keep patients away from polluted urban, industrial
and domestic environments.
DISCUSSION
The present analysis suggests that air pollution result-
ing from coal combustion directly affects TB incidence.
Although the decrease in TB observed in the West can
be explained by socio-economic factors, the use of
anti-tuberculosis drugs and/or the BCG vaccine, the
coincidence with coal consumption is nothing short
of troubling.
In China, HIV prevalence is low48 and therefore
cannot explain the increase in TB in the last 20 years.
The correlation may, however, reflect a direct associ-
ation between coal consumption and TB incidence. In
the case of China, notification rate instead of TB inci-
dence was used (Figure 4). It should be added that a
better health system in a wealthier state can result in
increased TB notifications.
Generally speaking, the evidence presented in the
present paper should be regarded with caution given
the inherent limitations of the ecological approach.
The correlations are prone to the ecological fallacy,
because no confounders can be tested. For example,
in Canada, the overall improvement of the health sys-
tem after WWII, concomitant with the electrification
of cities, could explain a global decrease in TB. Nev-
ertheless, it is an hypothesis to be investigated, as bio-
mass cooking fuel7 and cigarettes9 have been identi-
Air pollution caused by coal impacts on tuberculosis 729
fied as risk factors for TB, and social factors may not
convincingly explain the aetiology of the disease.5
The US Environmental Protection Agency esti-
mated US coal emissions in 2000 by fuel combustion
electric utilities for the principal pollutants: sulphur
dioxide (SO2) 10 723 thousand short tons (tst), nitro-
gen oxides (NOx) 4573 tst, CO 234 tst, PM 10 m
(PM10) 242 tst, PM 2.5 m (PM2.5) 116 tst, volatile
organic compounds (VOC) 30 tst, and lead (Pb) 56
short tons.49
The prime pollutant in coal is SO2, followed by
NOx. CO is substantially less, but partial combustion
of coal is known to produce large amounts of CO,
and it has been suggested that US tabulated data for
CO emissions may be underestimated.50 Ash from coal
combustion, or PM, is composed primarily of oxides
of silicon, aluminium and iron. On the whole, coal is
considered to be a very polluting fuel.
Bygone coal furnace technologies are likely to have
had an impact on indoor air pollutants and CO levels.
They were not nearly as efficient as they are today, and
therefore resulted in partial combustion of coal, indoor
build-up of CO and heavy PM emissions. Today in
China, low-grade coal is often burned in cheap, poorly
vented or even unvented stoves, resulting in various ill-
nesses.33 In addition, keeping stockpiles of moist lignite
coal near a source of heat will result in CO leakage, a
process reminiscent of industrial coal gasification.
Considering the present hypothesis, the impact of
coal on TB disease should transpose into a specific
mechanism in vivo. A closer look at the physiological
effects of CO and the biology of TB suggests a plau-
sible mechanism whereby coal fumes shut down the
immune response of the host.
Termed the silent killer, CO is an odourless, taste-
less and colourless gas with a well-known tendency to
build up indoors. Identified in recent years as an intra-
cellular signalling molecule, CO is generated endoge-
nously by haeme oxygenase enzymes (HOs). The IL-10
anti-inflammatory pathway requires the activation of
haeme oxygenase-1 (HO-1), which degrades the haeme
of haemoglobin, yielding CO, biliverdin and iron as
by-products.51 Anti-inflammatory action was found
to be mediated by CO, as its removal blocks the effects
of IL-10 and HO-1.52 A major effect of CO is the down-
regulation of tumour necrosis factor-alpha (TNF-).
TNF- plays a central role in the formation of
granulomata and containment of TB.53,54 Inhibition
of TNF- by drugs on trial has been linked to reacti-
vation of TB.55,56 In addition, macrophages are of par-
amount importance in the defence against M. tuber-
culosis. Growth inhibition is mostly mediated by
nitrogen oxide NO species57–60 promoted by TNF-.
Interferon-gamma (IFN-) and TNF- are central
to the process, as they induce macrophage activation,
inducible nitric oxide synthase (iNOS) expression and
NO generation.60,61
On the other hand, M. tuberculosis encodes trun-
730 The International Journal of Tuberculosis and Lung Disease
cated haemoglobins (trHbs) whose function remains
unclear. TrHbs have been associated with protection
against nitrosative damage by virtue of their NO di-
oxygenase activity, utilisation of CO as a carbon
source, O2/NO sensing, redox reactions, O2 delivery
in hypoxic conditions and long term ligand/substrate
storage.62–66 M. tuberculosis trHb has, along its haeme
group, an a-polar tunnel system through the protein
matrix endowed with more than one entry/exit points
and is able to store diatomic molecules in its cavity.67
The model shown in Figure 8 details a mechanism
whereby M. tuberculosis responds to macrophage ni-
trosative attack with the release of CO stored in trHb
molecules, followed by suppression of immunity, a
model termed ‘CO-doping’. In addition to actively
protecting the bacilli through NO oxidation by trHb
nitric dioxygenase activity, CO-doping functions to
induce an IL-10-like response. As a result, a decline in
TNF- production, macrophage activation and cell-
mediated immunity ensues. In either case, irrespective
of the capacity of M. tuberculosis to actively trigger
the IL-10 pathway, elevated ambient levels of CO re-
sult in lower levels of TNF- through activation of
the IL-10 pathway.
It is noteworthy to add that polymorphism in the
slc11a1 (formerly Nramp1) gene has been identified
as a factor of resistance to TB in mice.68 slc11a1 is
thought to be an iron transporter localised to mem-
branes of phagocytes in macrophages. Polymorphisms
in IL-10 and TNF- have also been linked to pulmo-
nary TB.69 The precise role of Slc11a1 in TB resis-
tance remains to be elucidated: however, it has been
Figure 8 CO-doping model in the reactivation of latent tuber-
culosis. Release to macrophages of CO stored in trHb of M.
tuberculosis, mimicking the IL-10 anti-inflammatory pathway,
results in inhibition of gene transcription, TNF- production and
leads to reduced macrophage activation, nitric oxide produc-
tion and cell-mediated immunity. CO  carbon monoxide; LPS 
lipopolysaccharide; TLR  toll-like receptor; trHb  truncated
haemoglobins; TNF  tumour necrosis factor; NO  nitric ox-
ide; HO  haeme oxygenase; IL-10  interleukin-10; iNOS 
inducible nitric oxide synthase.
suggested that Slc11a1 influences susceptibility to TB
by regulation of IL-10 because a polymorphism in
Slc11a1 was associated not only with TB, but also with
a higher endotoxin-induced production of IL-10.70 In
the scope of the CO-doping model, modification of
the iron pool by Slc11a1 serves to desensitise the IL-10
pathway in lung macrophages or to capture available
iron used in trHb haeme, thereby impacting on the
anti-inflammatory activity of CO and/or the ability of
M. tuberculosis to feed macrophages with CO.
Another issue yet to be considered is the link be-
tween PM from coal combustion and the activation
of the IL-10 pathway in fibrotic lungs. It is known
that miners in coal and gold mines are more prone to
TB as a result of silicosis.10,11 It has been shown that,
in mice, the IL-10 pathway is activated in the course
of fibrosis of the lung,71,72 but not in rats.
Coal ash differs from wood ash, which could ex-
plain an enhanced activation of TB by coal as opposed
to wood; TB existed before industrialisation, but pre-
sumably to a lesser extent. Logically, the amount of
CO produced by the combustion of coal should be
greater than that produced by the combustion of wood.
In combustion experiments for different fuels in a lab-
oratory system designed to mimic the fire pit used in
Xuan Wei County in China, emissions of anthracite,
bituminous coal and pine wood were compared.73
These results show that the total PM emissions for bi-
tuminous coal were 5.6-fold those of pine wood,
whereas there were only 1.6-fold more CO emissions
for bituminous coal than for pine wood. However, in
China, the use of cheap grades of coal such as moist
lignite may result in worse figures for CO.33
In conclusion, coal, the workhorse of progress,
seems an inevitable step in the emergence of an indus-
trialised country. Thus, in the light of the present hy-
pothesis, the white plague will rise and vanish along
the path of entry into modernity; the disease will fluc-
tuate in the population in a complex interplay of fac-
tors: domestic and industrial use of various types of
coal, combustion technologies, the number and gender
of workers in manufactures, and the source of mechan-
ical drive. Finally, as electrification progresses and pov-
erty plummets, TB regresses to a bad memory—until
AIDS brings it back on topic. Jean and René Dubos,
in their book The White Plague, called TB a disease of
‘incomplete civilization’.74 It could also be said that
TB is a disease of incomplete industrialisation.
Although micro-organisms are causative agents of
many illnesses, environmental health factors are not
to be underestimated in the aetiology of the disease.
The present evidence should prompt investigators to
study the effects of air pollution, notably coal, in re-
gard to TB. An epidemiological approach that can
take into account confounders could substantiate the
hypothesis. Furthermore, the impact of various air
pollutants on the outcome of TB should be investi-
gated in animal models.

Acknowledgements
I would like to thank Dr D Menzies for critical review of the manu-
script. I am also grateful to Dr T Shimao for information on histor-
ical facts pertaining to TB in Japan.
References
1 Havlir D V, Barnes P F. Tuberculosis in patients with human
immunodeficiency virus infection. N Engl J Med 1999; 340:
367–373.
2 World Health Organization. WHO report 2006. Global tuber-
culosis control: surveillance, planning, financing. WHO/HTM/
TB/2006.362. Geneva, Switzerland: WHO, 2006.
3 Grigg E R N. The arcana of tuberculosis, with a brief epidemi-
ologic history of the disease in the USA. Am Rev Tuberc Pulm
Dis 1968; 78: 151–172.
4 Shimao T. Mortality rate of pulmonary TB in the late 19th cen-
tury. Kekkaku 2003; 78: 633–636.
5 Davies R P O, Tocque K, Bellis M A, Remmington T, Davies
P D O. Historical declines in tuberculosis in England and Wales:
improving social conditions or natural selection? Vesalius 1999;
5: 25–29.
6 Marie-Davy H. Revue d’hygiène 1911; 33: 714.
7 Mishra V K, Retherford R D, Smith K R. Biomass cooking fuels
and prevalence of tuberculosis in India. Int J Infect Dis 1999; 3:
119–129.
8 Gupta B N, Mathur N. A study of household environmental
risk factors pertaining to respiratory diseases. Energy Environ
Monitor 1997; 13: 61–67.
9 Gajalakshmi V, Peto R, Kanaka T S, Jha P. Smoking and mor-
tality from tuberculosis and other diseases in India: retrospec-
tive study of 43 000 adult male deaths and 35 000 controls.
Lancet 2003; 352: 507–515.
10 Rockette H E. Cause-specific mortality of coal miners. J Occup
Med 1977; 17: 795–801.
11 Godfrey-Faussett P, Sonnenberg P, Shearer S C, et al. Tubercu-
losis control and molecular epidemiology in a South African
gold-mining community. Lancet 2000; 356: 1066–1071.
12 Statistics Canada. Historical statistics of Canada, catalog 11-
516-XIE. Ottawa, Canada: Statistics Canada, 1983.
13 Malissard P. La longue controverse de la vaccination antituber-
culeuse au Canada : le bacille Calmette-Guérin (BCG), 1925–
1975. Canadian Bulletin of Medical History 1998; 15: 87–128.
14 United States Department of Energy. Annual energy review
2005. Washington, DC, USA: Energy Information Administra-
tion, DOE, July 2006.
15 Centers for Disease Control and Prevention. Reported tubercu-
losis in the United States, 2005. Atlanta, GA, USA: US Depart-
ment of Health and Human Services, CDC, September 2006.
16 National Bureau of Statistics. China statistical yearbook 2003.
Beijing, China: National Bureau of Statistics, 2003a.
17 World Health Organization. WHO report 2005. Country pro-
file of China. Geneva, Switzerland: WHO, 2005.
18 Ravallion M, Chen S. China’s (uneven) progress against pov-
erty. World Bank Policy Research Working Paper No. 3408.
Washington, DC, USA: World Bank, 2004.
19 Rygg N. Statistike oversigter. Oslo, Norway: Statistik sentral-
byrå, 1914.
20 Norges offisielle statistikk Fjerde Række. Industristatistikk Nr.
99–NOS X. 44 (1895–1939). Oslo, Norway: Statistik cen-
tralbyrå, annual publications.
21 Wood S R. A contribution to the history of tuberculosis and lep-
rosy in 19th century Norway. J Roy Soc Med 1991; 84: 428–431.
22 Bouteville R. L’éclairage public à Paris. Paris, France: Léon Ey-
rolles, 1925.
23 Barnes D S. The making of a social disease: tuberculosis in
nineteenth century France. Berkeley, CA, USA: University of
California Press, 1995.
Air pollution caused by coal impacts on tuberculosis 731
24 Waldron H A. Occupational health during the Second World
War: hope deferred or hope abandoned? Med Hist 1997; 41:
197–212.
25 Tuberculosis—respiratory and non-respiratory notifications,
England and Wales, 1913–2004: Communicable Disease Sur-
veillance Centre (CDSC). London, UK: Health Protection
Agency, 2005.
26 Sepkowitz K A. Tuberculosis and the health care worker: a his-
torical perspective. Ann Intern Med 1994; 120: 71–79.
27 Fine P E M. Bacille Calmette-Guérin vaccines: a rough guide.
Clin Infect Dis 1995; 20: 11–14.
28 Rodrigues L C, Diwan V K, Wheeler J G. Protective effect of
BCG against tuberculous meningitis and miliary tuberculosis:
a meta-analysis. Int J Epidemiol 1993; 22: 1154–1158.
29 Colditz G A, Berkey C S, Mosteller F, et al. The efficacy of ba-
cillus Calmette-Guerin vaccination of newborns and infants in
the prevention of tuberculosis: meta-analyses of the published
literature. Pediatrics 1995; 96: 29–35.
30 Colditz G A, Brewer T F, Berkey C S, et al. Efficacy of BCG
vaccine in the prevention of tuberculosis. Meta-analysis of the
published literature. JAMA 1994; 271: 698–702.
31 Grzybowski S, Allen E A. Tuberculosis: 2. History of the dis-
ease in Canada. CMAJ 1999; 160: 1025–1028.
32 Landes D S. The unbound Prometheus: technological change
and industrial development in Western Europe from 1750 to the
present. Cambridge, UK: Cambridge University Press, 1969.
33 Finkelman R B, Belkin H E, Zheng B. Health impacts of do-
mestic coal use in China. Proc Natl Acad Sci USA 1999; 96:
3427–3431.
34 Smith K R, Liu Y. Epidemiology of lung cancer. New York, NY,
USA: Marcel Dekker, 1994.
35 Murray J F. A century of tuberculosis. Am J Respir Crit Care
Med 2004; 169: 1181–1186.
36 Holmberg S D. The rise of tuberculosis in America before
1820. Am Rev Respir Dis 1990; 142: 1228–1232.
37 Venneslan C. Labour markets and recovery of the great depres-
sion in Norway. XIV International Economic History Congress.
Helsinki, Finland: International Economic History Association,
21–25 August 2006, Session 10.
38 Ma D. Between cottage and factory: the evolution of Chinese
and Japanese silk-reeling industries in the latter half of the 19th
century. Asia Pacific Economy 2005; 10: 195–213.
39 Sugiyama S, Yamada I. The problem of fuel: an economic his-
tory of the environment with reference to the Nagano silk reel-
ing industry. Shakai keizai shigaku 1999; 65: 3–24. [Japanese]
40 Wakker B. High temperature corrosion in gasifiers. Materials
Res 2004; 7: 53–59.
41 Tristan F. Promenades dans Londres, 1842, l’aristocratie et les
prolétaires anglais. Paris, France: Indigo & Côté-femmes, 2001.
42 Winter C J, Nitch J. Hydrogen as an energy carrier. New York,
NY, USA: Springer Verlag, 1988.
43 Hastings C. Industrial disease and industrial hygiene. Can Med
Assoc J 1914; 4: 782–792.
44 McCarthy O R. The key to the sanatoria. J R Soc Med 2001;
94: 413–417.
45 Philip R W. Further experiences in the open-air treatment of
phthisis. Practitioner 1899; 10: 11–21.
46 Burton-Fanning F W. The open-air treatment of phthisis in
England. Lancet 1898; i: 630–633, 712–616, 854–637.
47 Bardswell N D, Thompson J H R. Mortality after sanatorium
treatment. Med Res Committee Spec Rep 1919; 33: 1–112.
48 Joint United Nations Programme on HIV/AIDS. Report on the
global AIDS epidemic. Geneva, Switzerland: UNAIDS, 2006.
49 United States Environmental Protection Agency. National air
quality and emissions trends report. Research Triangle Park,
NC, USA: US EPA, 2004.
50 Nicks D K J, Holloway J S, Ryerson T B, et al. Fossil-fuelled
power plants as a source of atmospheric carbon monoxide.
J Environ Monit 2003; 5: 35–39.
732 The International Journal of Tuberculosis and Lung Disease
51 Maines M D. The heme oxygenase system: update 2005.
Antioxid Redox Signal 2005; 7: 1761–1766.
52 Otterbein L E, Bach F H, Alam J, et al. Carbon monoxide has
anti-inflammatory effects involving the mitogen-activated pro-
tein kinase pathway. Nat Med 2000; 6: 422–428.
53 Mohan V P, Scanga C A, Yu K, et al. Effects of tumour necrosis
factor alpha on host immune response in chronic persistent
tuberculosis: possible role for limiting pathology. Infect Immun
2001; 69: 1847–1855.
54 Van Deventer S J H. Transmembrane TNF-, induction of apo-
ptosis, and the efficacy of TNF-targeting therapies in Crohn’s
disease. Gastroenterol 2001; 121: 1242–1246.
55 Keane J, Gershon S, Wise R P, et al. Tuberculosis associated
with infliximab, a tumour necrosis factor alpha-neutralizing
agent. N Engl J Med 2001; 345: 1098–1104.
56 Wagner T E, Huseby E S, Huseby J S. Exacerbation of Myco-
bacterium tuberculosis enteritis masquerading as Crohn’s
disease after treatment with a tumour necrosis factor-alpha in-
hibitor. Am J Med 2002; 112: 67–69.
57 Ehrt S, Schnappinger D, Bekiranov S, et al. Re-programming of
the macrophage transcriptome in response to interferon-
gamma and Mycobacterium tuberculosis: signalling roles of ni-
tric oxide synthase-2 and phagocyte oxidase. J Exp Med 2001;
194: 1123–1140.
58 MacMicking J D, North R J, LaCourse R, et al. Identification
of nitric oxide synthase as a protective locus against tuberculo-
sis. Proc Natl Acad Sci USA 1997; 94: 5243–5248.
59 Chan J, Tanaka K, Carroll D, et al. Effects of nitric oxide syn-
thase inhibitors on murine infection with Mycobacterium
tuberculosis. Infect Immun 1995; 63: 736–740.
60 Chan J, Xing Y, Magliozzo R S, et al. Killing of virulent Myco-
bacterium tuberculosis by reactive nitrogen intermediates pro-
duced by activated murine macrophages. J Exp Med 1992; 175:
1111–1122.
61 Manca C, Tsenova L, Barry C E, et al. Mycobacterium tuber-
culosis CDC1551 induces a more vigorous host response in
vivo and in vitro, but is not more virulent than other clinical
isolates. J Immunol 1999; 162: 6740–6746.
62 Ouellet H, Ouellet Y, Richard C, et al. Truncated haemoglobin
On considère généralement que la tuberculose (TB) va
de pair avec l’industrialisation et l’urbanisation. Avec un
pic dans les années 1800 et une lente régression ultérieure,
la maladie a décliné fortement en Occident après la deu-
xième guerre mondiale. La TB fait un retour dans les 20
dernières années dans les pays en développement comme
la Chine et l’Inde. Comme les seules conditions socio-
économiques ne peuvent pas expliquer la connexion entre
l’industrialisation et la TB, des facteurs étiologiques de
La tuberculosis (TB) se considera generalmente vincu-
lada con la industrialización y la urbanización. La enfer-
medad presentó el pico de frecuencia más alto en el siglo
XIX, declinó luego lentamente y mostró una dismi-
nución drástica después de la segunda guerra mundial,
en Occidente. Durante los últimos 20 años, la TB ha sur-
gido de nuevo en países en desarrollo como China e
India. Puesto que las condiciones socio-económicas por
sí solas no explican la correlación entre industrialización
y TB, deben determinarse los demás factores que con-
HbN protects Mycobacterium bovis from nitric oxide. Proc
Natl Acad Sci USA 2002; 99: 5902–5907.
63 Park S W, Hwang E H, Park H, et al. Growth of mycobacteria
on carbon monoxide and methanol. J Bacteriol 2003; 185:
142–147.
64 Wittenberg J B, Bolognesi M, Wittenberg B A, et al. Truncated
haemoglobins: a new family of haemoglobins widely distrib-
uted in bacteria, unicellular eukaryotes and plants. J Biol Chem
2002; 277: 871–874.
65 Gardner P R, Gardner A M, Martin L A, et al. Nitric-oxide di-
oxygenase activity and function of flavohaemoglobins. Sensi-
tivity to nitric oxide and carbon monoxide inhibition. J Biol
Chem 2000; 275: 31581–31587.
66 Mukai M, Mills C E, Poole R K, et al. Flavohaemoglobin, a
globin with a peroxidase-like catalytic site. J Biol Chem 2001;
276: 7272–7277.
67 Milani M, Pesce A, Ouellet Y, et al. Heme-ligand tunneling in
group I truncated haemoglobins. J Biol Chem 2004; 279: 21520–
21525.
68 Kramnik I, Dietrich W F, Demant P, et al. Genetic control of re-
sistance to experimental infection with virulent Mycobacterium
tuberculosis. Proc Natl Acad Sci USA 2000; 97: 8560–8565.
69 Delgado J C, Baena A, Thim S, et al. Ethnic-specific genetic
associations with pulmonary tuberculosis. J Infect Dis 2002;
186: 1463–1468.
70 Awomoyi A A, Marchant A, Howson J M M, et al. Interleukin-
10, polymorphism in SLC11A1 (formerly NRAMP1), and sus-
ceptibility to tuberculosis. J Infect Dis 2002; 186: 1808–1814.
71 Barbarin V, Nihoul A, Misson P, et al. The role of pro- and
anti-inflammatory responses in silica-induced lung fibrosis.
Respir Res 2005; 6: 1–13.
72 Huaux F, Louahed J, Hudspith B, et al. Role of interleukin-10
in the lung response to silica in mice. Am J Respiratory Cell
Mol Biol 1998; 31: 78–85.
73 Tian L. Coal combustion emissions and lung cancer in Xuan
Wei, China. Environmental Sciences Dissertation. Berkeley,
CA: University of California, Berkeley, 2005: p 221.
74 Dubos R, Dubos J. The white plague: tuberculosis, man, and so-
ciety. New Brunwsick, NJ, USA: Rutgers University Press, 1987.
RÉSUMÉ
la maladie doivent encore être déterminés. On met en
corrélation les statistiques historiques de consommation
de charbon et de maladie TB au Canada, aux USA et en
Chine. On élabore une hypothèse liant la TB et la pollu-
tion de l’air dans le contexte de l’industrialisation. Un
modèle est proposé dans lequel la stimulation de la cascade
IL-10 par le monoxyde de carbone dans les macrophages
pulmonaires favorise la réactivation de Mycobacterium
tuberculosis.
RESUMEN
tribuyen a la etiología de la enfermedad. Las estadísticas
históricas de consumo de carbón presentan una correla-
ción con la incidencia de tuberculosis en Canadá, los
Estados Unidos y China. Se plantea una hipótesis que
asocia la TB con la contaminación atmosférica en el
contexto de la industrialización. Se propone un modelo
según el cual la activación de la cascada de la inter-
leucina 10 por el monóxido de carbono en los macrófa-
gos del pulmón promueve la reactivación de Mycobacte-
rium tuberculosis.