Pipe smoking and cancers of the upper digestive tract
Giorgia Randi1,2*, Lorenza Scotti1, Cristina Bosetti1, Renato Talamini3, Eva Negri1, Fabio Levi4,
Silvia Franceschi5and Carlo La Vecchia1,2
1Department of Epidemiology, Istituto di Ricerche Farmacologiche ‘‘Mario Negri’’, Milan, Italy
2Istituto di Statistica Medica e Biometria ‘‘G.A. Maccacaro’’, Universita ` degli Studi di Milano, Milan, Italy
3Servizio di Epidemiologia e Biostatistica, Centro di Riferimento Oncologico, Aviano, Italy
4Unit? e d’? epid? emiologie du cancer, Institut Universitaire de M? edicine Sociale et Pr? eventive, Universite ´ de Lausanne,
5International Agency for Research on Cancer, Lyon, France
Pipe smoking has been related to the risk of cancers of the upper
digestive and respiratory tract, but quantification of the risk for
exclusive pipe smokers is still limited. To analyse the association
between exclusive pipe smoking and cancers of the upper digestive
tract, we used data from a series of case–control studies conducted
in Italy and Switzerland between 1984 and 1999. After excluding
cigarette and cigar smokers, 41 male oral and pharyngeal cancer
cases, 52 male oesophageal cancer cases and 1,032 male controls
were included in the present analysis. Odds ratios (OR) of cancers
were estimated by the mean of unconditional multivariate logistic
regression, including terms for age, study centre, education, body
mass index, and alcohol drinking. Compared to never smokers,
exclusive pipe smokers had an OR of 8.7 [95% confidence inter-
vals (CI): 4.0–18.9] of all upper digestive tract cancers. The OR
was 12.6 for oral and pharyngeal and 7.2 for oesophageal cancer.
Pipe smokers who were also heavy alcohol drinkers had an OR of
38.8 (95% CI: 13.6–110.9) as compared to never smokers and light
drinkers. Thus, pipe smoking and heavy alcohol drinking appears
to interact at least on a multiplicative model.
' 2007 Wiley-Liss, Inc.
Key words: digestive cancer; pipe smoking; case–control study; risk
Pipe smoking has been related to the risk of cancers of the upper
digestive and respiratory tracts,1,2but quantification of the risk for
exclusive pipe smokers is still limited. The American Cancer Soci-
ety Cancer Prevention Study II (ACS-CPS II)2reported relative
risk (RR) of oral and pharyngeal cancers of 3.9, based on 15
deaths, and of oesophageal cancer of 2.4, based on 20 deaths, for
current pipe smokers. Another cohort of 25,129 Swedish men en-
rolled in 1963 and followed-up to 1979 gave RRs of 1.4 for
cancers of the oral cavity, pharynx and larynx (based on 3 deaths)
and of 3.6 for cancer of the oesophagus (based on 6 deaths).3
Other cohort studies from the USA4,5reported similar RR for oral,
pharyngeal and oesophageal cancers, again based on small num-
bers. Finally, in a study including 104 case/control pairs from Bei-
jing, China, the multivariate odds ratio (OR) of oral cancer among
exclusive pipe smoking was 5.7 in males and 4.9 in females.6
To provide further quantification of the risk of upper digestive
tract neoplasms for exclusive pipe smokers, we analysed data
from a network of case–control studies conducted in Italy and
Material and methods
The present analysis is based on data from 3 hospital-based
case–control studies of cancers of the oral cavity and pharynx, and
3 case–control studies of the cancer of the oesophagus, whose
methods have already been described.7–11Only male subjects
were included in the present analysis, since female cancers of the
upper digestive tract are rare, and the proportion of female pipe
smokers is negligible.
Two Italian case–control studies on oral and pharyngeal cancers
were conducted between 1984 and 1997 in the greater Milan area
and the provinces of Pordenone, Rome, and Latina on 961 inci-
dent, histologically confirmed male cases.7,8Another study on
oral and pharyngeal cancers was conducted between 1992 and
1997 in the Swiss Canton of Vaud and included 126 male cases.10
Two Italian studies on oesophageal cancer were conducted
between 1984 and 1998 in the greater Milan area and in the prov-
inces of Pordenone and Padua on 618 male cases.9Another study
on oesophageal cancer was conducted between 1992 and 1999 in
the Canton of Vaud on 82 cases.11Thus, 1,787 male cases (1,087
cases of oral and pharyngeal cancers and 700 of oesophageal can-
cer) aged between 25 and 82 years (median age 59 years) were
included in the original studies. In all case–control studies, con-
trols were men admitted to the same network of hospitals as cases
for a wide spectrum of acute, non-neoplastic conditions, neither
related to smoking and alcohol consumption nor to long-term diet
modifications. Overall, there were 3,829 nonoverlapping controls
aged between 25 and 85 years (median age 57 years). Of these 27%
were admitted for non-alcohol-related traumas, 28% for nontrau-
matic orthopaedic disorders, 28% for acute surgical conditions, and
17% for miscellaneous other illnesses, including eye, ear, nose,
throat, skin or dental disorders.
To analyze the association between pipe smoking and cancers of
upper digestive tract we considered data on exclusive pipe smokers
and never smokers, excluding cigarette and/or cigar smokers. Thus,
the present analyses are based on 41 cases (median age 60 years)
of oral and pharyngeal cancers, 52 cases (median age 62 years) of
oesophageal cancer, and 1,032 controls (median age 56 years).
Response rate was over 95% in Italy, and around 85% in Swit-
zerland for both cases and controls. Trained interviewers inter-
viewed cases and controls during their hospital stay using similar
structured questionnaires. These included information on socio-
demographic characteristics, anthropometric measures, lifestyle
habits, such as tobacco smoking and alcohol drinking, a problem-
oriented personal medical history, and family history of cancer.
The section on smoking included question on smoking status
(never, current, former smokers), daily number of cigarettes or
cigars and grams of tobacco for pipe smoked, age at starting, dura-
tion of the habit and, for former smokers, age at smoking cessa-
tion. Former smokers were subjects who had abstained from any
type of smoking for at least 12 months.
ORs and the corresponding 95% confidence intervals (CIs)
were derived from unconditional multivariate logistic regression
models, fitted by the method of maximum likelihood,12including
terms for age (quinquennia), study centre, years of education (<7,
Grant sponsors: Italian Association for Cancer Research, Italian and
Swiss Leagues Against Cancer; Grant sponsor: Swiss Foundation for
Research Against Cancer; Grant numbers: AKT 413 and 700; Grant spon-
sor: Italian Ministry of Research; Grant number: PRIN 2005.
*Correspondence to: Istituto di Ricerche Farmacologiche ‘‘Mario
Negri’’-Via Eritrea, 62-20157 Milano, Italy. Fax: 10039-02-33200231.
Received 22 December 2006; Accepted after revision 14 March 2007
Published online 13 July 2007 in Wiley InterScience (www.interscience.
Int. J. Cancer: 121, 2049–2051 (2007)
' 2007 Wiley-Liss, Inc.
Publication of the International Union Against Cancer
7–11, ?12 years), body mass index (<25, 25–30, ?30 kg/m2),
and alcohol drinking (<14, 14–21, 22–34, 35–48, >48 drinks per
The significance of the interaction for the combined effect of pipe
smoking and alcohol drinking was assessed by comparing the differ-
ences between the deviances of the models with and without the
interaction term to the v2distribution with 1 degree of freedom.
The distribution of 93 male cases of cancers of the upper diges-
tive tract (41 oral cavity and pharyngeal cancers and 52 oesopha-
geal cancer cases), and 1,032 male controls according to age, edu-
cation, body mass index, and alcohol consumption is reported in
Table I. Male cases were older, thinner and reported higher alco-
hol consumption than male control subjects.
Table II gives the distribution of cases of cancers of the upper di-
gestive tract, and male controls according to exclusive pipe smoking,
and the corresponding OR. Compared to never smokers, the OR of
all upper digestive tract cancers was 8.7 for exclusive pipe smokers
and was 8.1 for current smokers only. The ORs were 7.9 for ?10 g
of tobacco per day and 9.1 for >10 g of tobacco per day, correspond-
ing to more than 3 pipes per day. The OR of pipe smokers was 12.6
for oral cavity and pharyngeal, and 7.2 for oesophageal cancer.
Table III gives the combined effect of exclusive pipe smoking
and alcohol consumption. Compared to never smokers and light
drinkers (?21 drinks per week), the ORs were 2.3 for never smokers
and heavy drinkers (>21 drinks per week), 4.1 for exclusive pipe
smokers and light drinkers and 38.8 for exclusive pipe smokers and
heavy drinkers. The interaction between alcohol and pipe smoking
was of borderline significance (p-value of interaction 5 0.05).
The data on upper digestive cancer risk in relation to pipe smok-
ing are much fewer than those available for cigarette smoking,
given the smaller proportion of people who exclusively smoke
pipe,13and the difficulties of adjusting for cigarettes and other
types of tobacco smoking in subjects who smoke pipes and other
tobacco products. From official national tobacco sales from 1993
to 2004 (ISTAT), pipe tobacco in Italy accounted for only about
0.4% of all tobacco smoking sales and exclusive pipe smoking
likely accounts for a much smaller proportion.
Despite being one of the largest available dataset, also this
study included only 18 cancer cases and 24 controls of exclusively
pipe smokers, thus making risk estimates subject to considerable
random variation, and hampering analysis of subgroups.
The almost 9-fold increased risk for exclusive pipe smokers
observed in this study is apparently higher than the RRs found in a
few cohort studies conducted in the USA and northern Europe.2–4
The higher risk observed in this study may be due to greater valid-
ity of information on pipe smoking, particularly in the short term
before cancer occurrence, since at least part of subjects classified
as current smokers at interview in cohort studies had stopped
smoking at cancer occurrence, and should therefore be considered
as former smokers.14The lower confidence limit of our estimate
is, however, around 4, i.e. compatible with most previous
reports.2–4Thus, the apparent discrepancies may be due to the role
of chance, as well as, at least in part, to the effect of recall bias in
cases and controls.
Pipe smoke is inhaled less than cigarette one, and hence is asso-
ciated with lower excess risk of lung cancer, as well as of pancre-
atic cancer, bladder cancer and of most other tobacco-related neo-
plasms.2,3,13However, the epithelia of the oral cavity and oeso-
phagus are in direct contact with pipe smoke, and this explains the
substantial excess cancer risk for these sites. Direct contact with
tobacco carcinogens may also explain the greater RR for oral and
pharyngeal than for oesophageal cancer observed in this as well as
in another study.2Similar line of reasoning applies to cigar smok-
ing, which has been associated to a 7-fold increased risk of upper
digestive tract cancer in our data.15
The data of our analysis were derived from hospital-based
case–control studies, and the use of hospital controls to analyze
smoking-related risk can be subject to bias. However, the preva-
lence of smoking in the control group was comparable to that of
the 1995 Italian National Multipurpose Survey16and to that of
Swiss Surveys.17,18Moreover, cases and controls were inter-
viewed in the same hospitals and came from the same geographic
area, the participation was almost complete, and we excluded
from the control group patients admitted to hospital for chronic
conditions, particularly those associated with smoking or alcohol
drinking, or diseases related to other known or likely risk factors
for cancers of the upper digestive tract, thus providing further
reassurance against potential information bias. Moreover, the in-
formation on cigarette smoking was satisfactorily reproducible.19
Risk estimates were adjusted for major covariates that could affect
TABLE I – DISTRIBUTION OF 93 MALE CASES OF CANCERS OF THE UPPER DIGESTIVE TRACT, AND 1,032
MALE CONTROLS WHO HAVE NEVER SMOKED CIGARETTES OR CIGARS, ACCORDING TO AGE, EDUCATION,
BODY MASS INDEX AND ALCOHOL CONSUMPTION. ITALY AND SWITZERLAND, 1984–1999
Oral cavity and
No.% No.% No.%
Body mass index (kg/m2)1
Alcohol drinking (drinks/week)1
1The sum does not add up to the total because of some missing values.
RANDI ET AL.
the association between pipe smoking and risk of upper digestive Download full-text
tract cancers, such as alcohol consumption, body mass index, and
education as a proxy indication of socioeconomic status. Addi-
tional allowance for measures of vegetable and fruit consumption,
which have been inversely related to risk of the upper digestive
tract cancers, did not materially modify any of the estimates [fur-
ther adjusted OR for ever exclusive pipe smokers vs. never smok-
ers was 10.52 (95% CI: 4.29–25.78)].
Despite the low number of exclusive pipe smokers and conse-
quently the wide CI of each single estimate, we were able to ana-
lyse the effect of the amount of pipe tobacco smoked, and the
combined effect of pipe smoking and heavy alcohol drinking. Sub-
jects exposed to both factors had an about 40-fold excess risk as
compared to never smokers and light drinkers. The borderline sig-
nificance of the interaction term indicates that, if anything, the 2
factors act, at least, on a multiplicative model on the risk of upper
digestive tract neoplasms.
The authors thank Mrs. C. Pasche and Mrs. F. Lucchini for
Swiss data collection and validation, and Mrs. I. Garimoldi for
1. Boffetta P, Pershagen G, Jockel KH, Forastiere F, Gaborieau V, Hein-
rich J, Jahn I, Kreuzer M, Merletti F, Nyberg F, Rosch F, Simonato L.
Cigar and pipe smoking and lung cancer risk: a multicenter study
from Europe. J Natl Cancer Inst 1999;91:697–701.
Henley SJ, Thun MJ, Chao A, Calle EE. Association between exclu-
sive pipe smoking and mortality from cancer and other diseases. J
Natl Cancer Inst 2004;96:853–61.
Carstensen JM, Pershagen G, Eklund G. Mortality in relation to ciga-
rette and pipe smoking: 16 years’ observation of 25,000 Swedish
men. J Epidemiol Community Health 1987;41:166–72.
Hammond EC, Horn D. Smoking and death rates: report on forty-four
months of follow-up of 187,783 men. II. Death rates by cause. J Am
Med Assoc 1958;166:1294–308.
Hammond EC. Smoking in relation to the death rates of one million
men and women. Natl Cancer Inst Monogr 1966;19:127–204.
Zheng TZ, Boyle P, Hu HF, Duan J, Jiang PJ, Ma DQ, Shui LP, Niu
SR, MacMahon B. Tobacco smoking, alcohol consumption, and risk
of oral cancer: a case-control study in Beijing, People’s Republic of
China. Cancer Causes Control 1990;1:173–9.
Franceschi S, Talamini R, Barra S, Baron AE, Negri E, Bidoli E, Ser-
raino D, La Vecchia C. Smoking and drinking in relation to cancers of
the oral cavity, pharynx, larynx, and esophagus in northern Italy. Can-
cer Res 1990;50:6502–7.
Talamini R, La Vecchia C, Levi F, Conti E, Favero A, Franceschi S.
Cancer of the oral cavity and pharynx in nonsmokers who drink alco-
hol and in nondrinkers who smoke tobacco. J Natl Cancer Inst
Zambon P, Talamini R, La Vecchia C, Dal Maso L, Negri E, Tog-
nazzo S, Simonato L, Franceschi S. Smoking, type of alcoholic bever-
age and squamous-cell oesophageal cancer in northern Italy. Int J
10. Levi F, Pasche C, La Vecchia C, Lucchini F, Franceschi S, Monnier
P. Food groups and risk of oral and pharyngeal cancer. Int J Cancer
11. Levi F, Pasche C, Lucchini F, Bosetti C, Franceschi S, Monnier P, La
Vecchia C. Food groups and oesophageal cancer risk in Vaud, Swit-
zerland. Eur J Cancer Prev 2000;9:257–63.
12. Breslow NE, Day NE. Statistical methods in cancer research, vol. 1:
the analysis of case-control studies. Lyon, France: IARC, 1980. IARC
Sci Publ No. 32.
13. IARC. Tobacco smoke and involuntary smoking. IARC Monogr Eval
Carcinog Risks Hum 2004;83:1–1438.
14. Negri E, La Vecchia C, D’Avanzo B, Nobili A, La Malfa RG. Acute
myocardial infarction: association with time since stopping smoking
in Italy. GISSI-EFRIM Investigators. Gruppo Italiano per lo Studio
della Sopravvivenza nell’Infarto. Epidemiologia dei Fattori di Rischio
dell’Infarto Miocardico. J Epidemiol Community Health 1994;48:
15. La Vecchia C, Bosetti C, Negri E, Levi F, Franceschi S. Cigar smoking
and cancers of the upper digestive tract. J Natl Cancer Inst 1998;90:1670.
16. Pagano R, La Vecchia C, Decarli A. Smoking in Italy, 1995. Tumori
17. La Vecchia C, Levi F, Decarli A, Wietlisbach V, Negri E, Gutzwiller
F. Trends in smoking and lung cancer mortality in Switzerland. Prev
18. Levi F, Colombo P, La Vecchia C. Effects of new smoking regula-
tions. Ann Oncol 2006;17:1335.
19. D’ Avanzo B, La Vecchia C, Katsouyanni K, Negri E, Trichopoulos D.
Reliability of information on cigarette smoking and beverage consump-
tion provided by hospital controls. Epidemiology 1996;7:312–15.
TABLE II – DISTRIBUTION OF 93 MALE CASES OF CANCERS OF THE UPPER DIGESTIVE TRACT, AND 1,032
MALE CONTROLS WHO HAVE NEVER SMOKED CIGARETTES OR CIGARS, ACCORDING TO PIPE SMOKING, AND
CORRESPONDING ODDS RATIOS (OR) AND 95% CONFIDENCE INTERVALS (CI). ITALY AND SWITZERLAND,
Cancer site, exclusive pipe smoking
All upper digestive cancers
Grams of tobacco per day
24 8.67 (3.99–18.85)
1Estimates from multiple logistic regression models including terms for age, center, education, body
mass index, and alcohol consumption.–2Reference category.
TABLE III – ODDS RATIOS (OR)1OF CANCERS OF THE UPPER DIGESTIVE
TRACT, AND CORRESPONDING 95% CONFIDENCE INTERVALS (CI)
ACCORDING TO ALCOHOL CONSUMPTION AND EXCLUSIVE PIPE SMOK-
ING. ITALY AND SWITZERLAND, 1984–1999
Exclusive pipe smoking
OR (95% CI)
OR (95% CI)
4.12 (1.22–13.92)38.82 (13.58–110.94)3
1Estimates from multiple logistic regression models including terms
for age, center, education, and body mass index.–2Reference cate-
gory.–3p 5 0.05
PIPE SMOKING AND UPPER DIGESTIVE TRACT CANCERS