Cancer Causes and Control 15: 771–780, 2004.

771

2004 Kluwer Academic Publishers. Printed in the Netherlands.

Tobacco and alcohol consumption and the risk of non-Hodgkin lymphoma

Eleanor V Willett1,*, Alexandra G. Smith1, Gareth J. Dovey1, Gareth J. Morgan2, Jan Parker1 & Eve Roman1
1
Department of Health Sciences, Leukaemia Research Fund Epidemiology and Genetics Unit, Seebohm Rowntree
Building, University of York, York YO10 5DD, UK; 2Institute of Cancer Research, The Royal Marsden, Downs Road,
Sutton, Surrey, SM2 5PT, UK

Received 2 February 2004; accepted in revised form 27 April 2004

Key words: tobacco, alcohol, non-Hodgkin lymphoma, epidemiology, case–control study.

Abstract

Objective: The aim was to test whether non-Hodgkin lymphoma (NHL) is associated with smoking or alcohol.
Methods: A case–control study recruited NHL cases aged 18–64 in parts of England between 1998 and 2001. One
control was matched to each case on sex, date of birth and area of residence. Self-reported histories of tobacco and
alcohol consumption were collected during face-to-face interviews.
Results: Among 700 cases and 915 controls, no association of smoking with the risk of NHL was observed [odds
ratio (OR) ¼ 1.04, 95% confidence interval (CI): 0.85–1.28]. Risks were not raised with age started smoking, number
of years smoked, and number of years stopped smoking. Compared with persons who drank alcohol once or twice a
week, neither abstainers (OR ¼ 1.03, 95% CI: 0.64–1.67), nor consumers of alcohol one to five times a year
(OR ¼ 1.35, 95% CI: 0.95–1.93), one to two times a month (OR ¼ 1.20, 95% CI: 0.87–1.65), three to four times a
week (OR ¼ 0.82, 95% CI: 0.62–1.10), or most days (OR ¼ 0.94, 95% CI: 0.70–1.25) increased their risk of
developing NHL. Average daily volume or high occasional alcohol consumption were not associated with NHL.
Conclusions: NHL was not associated with smoking or alcohol, but collaborative studies could further investigate
the risks of rarer WHO subtypes following these exposures.

Introduction gations have suggested associations between NHL and

Little is known of the causes of non-Hodgkin lymphoma
(NHL). A small proportion of rare NHL subtypes are
caused by severe immunosuppression arising from a few
rare inherited disorders of the immune system, long-
term immunosuppressive drug therapy, and certain
viruses such as human immunodeficiency virus (HIV)
and Epstein–Barr virus, for instance. Not only do
tobacco smoke and alcoholic beverages contain or
metabolize to carcinogenic compounds, but exposure
to either tobacco smoke or alcohol can lead to suppres-
sion of the immune system [1–5]. Hence, the hypothesis
that tobacco smoke and alcohol are associated with
NHL seems plausible. Previous epidemiological investi-
* Address correspondence to: Eleanor V. Willett, Department of
Health Sciences, Leukaemia Research Fund Epidemiology and
Genetics Unit, Seebohm Rowntree Building, University of York,
York YO10 5DD UK. Ph.: 44(0)190 432 1892; Fax:
44(0)190 432 1320; E-mail: eleanor.willett@egu.york.ac.uk
tobacco smoke or alcohol consumption, while others
have not. The current report presents findings for
smoking and alcohol consumption from a large popu-
lation-based case–control study of lymphoma.
Materials and methods
The Leukaemia Research Fund’s case–control study of
lymphoma began on 1 January 1998 in the English
counties of North, East and West Yorkshire, the county
of Lancashire, and the district of South Lakeland, and
on 1 April 1998 in the Caradon district of Cornwall,
South Devon, Dorset and South Hampshire. Approx-
imately four million 18–64-year-olds live in this study
area (Figure 1). The study was conducted with the
ethical approval of the United Kingdom Multi-regional
Ethical Committee. Cases were all persons newly diag-
nosed with NHL or HL, aged 18–64, and normally
resident in the study area. Ascertainment was through
772
the Haematological Malignancy Diagnostic Service
(HMDS) at the Leeds General Infirmary [6] for cases
diagnosed in Yorkshire, or through weekly or monthly
visits by a trained investigator to haematological
departments of all hospitals in, and on the peripheries,
of the other study regions. Diagnoses, where patholog-
ical material was available, were reviewed centrally and
coded to the World Health Organisation Classification
[7]. Ascertainment of patients diagnosed with diffuse
large B-cell lymphoma (ICD-O3 9679–9684) or follicle
centre cell lymphoma (ICD-O3 9690–9698) ceased on 31
March 2001, while continuing for the rarer diagnoses of
marginal zone lymphoma (ICD-O3 9689 and 9699),
mantle cell lymphoma (ICD-O3 9673), T-cell lymphoma
(ICD-O3 9700–9719 and 9827), nodular lymphocyte
predominant HL (ICD-O3 9659), and classical HL
(ICD-O3 9650–9655 and 9663–9667). Patients were
ineligible if they had a previous diagnosis of lymphoma,
or if their current diagnosis of lymphoma was associated
with HIV. With the consent of the treating consultant,
eligible cases were asked to participate in the study as
soon as possible after their diagnosis, with the median
time from diagnosis to interview being 92 days (range
0–807 days). Interviews were not undertaken with any
patient who had insufficient command of English, or
was severely ill; further, no surrogate interviews were
attempted for patients who had died before they were
approached.
One control per case, individually matched on sex,
date of birth, and residence at age of diagnosis, was
randomly selected from general practice lists. The
majority of controls were selected from the general
practice where the case was registered; if the general
practitioner (GP) of the case refused, an adjacent
general practice was approached. Controls were eligible
under the same study criteria as cases. Following the
consent of their GP, potential controls were sent a letter
inviting them to participate. If no reply was received
within ten working days, contact was attempted via a
telephone call, followed by a subsequent letter of
invitation. If ten working days after posting the second
letter of invitation, no reply had been received, or if the
control had refused, the next eligible control was
approached. Controls were assigned a reference date
that corresponded to the exact age of their matched case
when diagnosed.
A total of 1209 lymphoma patients diagnosed between
the commencement of the study and 31 March 2001
were identified. Interviews were obtained for 912 (75%)
cases. Permission to interview was refused by 40 (3%)
consultants and 89 (7%) patients. No interviews were
undertaken for a further 168 (14%) cases who had either
died before they were approached, were severely ill, had
E.V. Willett et al.
insufficient command of English, or could not be traced.
Among the 912 participating patients, 872 (96%)
diagnoses were confirmed. GPs gave permission to
approach 1706 controls. Of the controls who were
successfully contacted, 919 (71%) agreed to be inter-
viewed and 371 (29%) refused. It is possible that the
majority of the 416 potential controls who could not be
contacted would have moved from the address supplied
by the GP. The present unmatched analysis is restricted
to 700 Caucasian cases with a confirmed diagnosis of
NHL and 915 Caucasian controls, including controls
matched to HL cases.
All participating subjects were asked to consent to
being interviewed, to give mouthwash and blood sam-
ples, and to allow access to their medical records; cases
were also requested to permit storage of their diagnostic
tissue samples for research under the study. Subjects
were asked to complete a pre-interview form that
requested details of their residential, occupational and
family medical histories. Subsequent interviews, con-
ducted face-to-face by trained personnel using a struc-
tured questionnaire collected information on
demographics, residential history, sun exposure, and
medical history as well as history of tobacco use and
alcohol consumption.
Exposure to tobacco was defined as ever smoking
cigarettes, cigars or a pipe at least once a day for a
minimum of six months. For each type of tobacco
smoked, the age or year started, the amount smoked per
day, and the age or year stopped were recorded for every
change in habit. A two-year latent period prior to
diagnosis/reference date was applied to the smoking
data. Alcohol consumption was defined as ever drinking
wine, spirits, beer or lager at least once a year in the
20 years preceding diagnosis/pseudo-diagnosis. At five,
ten and 20 years prior to diagnosis, the frequency and
the amount typically consumed on each occasion of beer
(half pints of beer, lager, stout, cider, or shandy, or
bottles of alcopops), wine (glasses of wine, sherry,
martini or fortified wine) and spirits (single units of
spirits or liqueurs) were recorded; each drink’s measure
approximated a unit of alcohol or 8 g of ethanol [8]. For
each time point, the average daily volume of alcohol was
estimated, by summing across the beverage types, the
product of the amount of each type per occasion and
the proportion of the year the type was drunk, where the
proportion was the mid-point, in days, of the frequency
period divided by 365. Pattern of occasional heavy
drinking was defined as consuming 8 units of alcohol or
more on at least one occasion a month [9]. While the
baseline group for the analysis of tobacco was never
smoked, comparisons of alcohol categories were relative
to the commonest exposure group, rather than the
Tobacco and alcohol consumption
alcohol abstainers, since there is the possibility that any
observed association may be artefact if the reason for
non-drinking is related to the outcome [10].
A deprivation indicator was created using the address
at diagnosis. For each address, the postcode was
validated against the Post Office Postcode Address File
using QuickAddressTM (v2.0) and matched to a 1991
census small area, or ‘‘enumeration district’’, via the
Pc2ed program available from the Manchester Com-
puting Service (MIMAS). Townsend scores were com-
puted for each enumeration district in England and
Wales using data from the 1991 census [11]. Categori-
zation of the resulting continuous variable provided a
coding scheme for the enumeration districts of the
addresses at diagnoses.
Odds ratios (OR) and 95% confidence intervals (CI)
for unmatched analyses, adjusted for sex, age and
region, were calculated using unconditional logistic
regression; risk estimates for matched analyses were
also computed using conditional logistic regression [12].
Tests for interaction were conducted using the likeli-
hood ratio test. All analyses were performed using Stata
[13].
Results
Among interviewed cases with a confirmed diagnosis of
NHL, 318 (45%) were diagnosed with diffuse large B-
cell lymphoma, 228 (33%) with follicle centre cell
lymphoma, 73 (10%) with marginal zone lymphoma,
28 (4%) with mantle cell lymphoma, 35 (5%) with T-cell
lymphoma, and 18 (3%) with NHL not otherwise
specified before 31 March 2001 (Table 1). A slightly
greater proportion of men were diagnosed with diffuse
large B-cell lymphoma (53%), mantle cell lymphoma
(82%), T-cell lymphoma (63%) and NHL not otherwise
specified (61%) than women but more women than men
were diagnosed with follicle centre cell lymphoma (55%)
and marginal zone lymphoma (52%). Few patients (4%)
were diagnosed before the age of 35, although diffuse
large B-cell lymphoma (7%) was more common in this
age group than follicle centre cell lymphoma (1%) or
any other subtype of NHL. No statistically significant
differences were observed between cases and unmatched
controls for sex, marital status, or deprivation. Cases
were more likely to be older (v2 ¼ 15.3, p ¼ 0.002) and to
have left school before the age of 16 years (v2 ¼ 6.71,
p ¼ 0.035) than unmatched controls. However, despite
the inclusion of controls matched to patients diagnosed
with HL, risk estimates for an unmatched analysis, as
reported here, were little different to those calculated in
a matched analysis.
773
Table 2 shows the number of cases and controls by
reported tobacco smoking status two years preceding
diagnosis/reference date. For all NHL cases and un-
matched controls, 433 (62%) cases and 549 (60%)
controls reported that they had smoked tobacco regu-
larly (OR ¼ 1.04, 95% CI: 0.85–1.28). Compared with
non-smokers, no effect was observed for either ex-
(OR ¼ 0.99, 95% CI: 0.78–1.26) or current smokers
(OR ¼ 1.10, 95% CI: 0.86–1.41). Similarly, age started
smoking, the number of years smoked, and the number
of years since stopped smoking did not alter the risk of
NHL. No trend was observed with the number of
cigarettes smoked per day (<15 cigarettes/day:
OR ¼ 1.12, 95% CI: 0.79–1.58 based on 81 cases and
104 controls; 15–24 cigarettes/day: OR ¼ 1.23, 95% CI:
0.90–1.68 based on 120 cases and 139 controls; ‡25
cigarettes/day: OR ¼ 1.16, 95% CI: 0.76–1.75 based on
55 cases and 70 controls). Risk estimates for diffuse large
B-cell lymphoma, follicle centre cell lymphoma, mantle
cell lymphoma, marginal zone lymphoma, T-cell lym-
phoma and NHL not otherwise specified are generally
equivalent to those observed for all NHLs combined.
Moreover, risks did not vary greatly when data were
stratified by age or sex (data not shown).
Table 3 shows the number of cases and controls by
alcohol consumption at five years prior to diagnosis/
reference date. Relative to persons who drank alcohol
once or twice a week, no statistically significant risk of
NHL was observed among current abstainers
(OR ¼ 1.03, 95% CI: 0.64–1.67), or among persons
who consumed alcohol one to five times a year
(OR ¼ 1.35, 95% CI: 0.95–1.93), one to two times a
month (OR ¼ 1.20, 95% CI: 0.87–1.65), three to four
times a week (OR ¼ 0.82, 95% CI: 0.62–1.10), or most
days (OR ¼ 0.94, 95% CI: 0.70–1.25). About 315 (45%)
cases and 359 (39%) controls drank, on average,
<1 unit of alcohol per day; greater mean daily volumes
of alcohol decreased, although not significantly, the risk
of NHL. No effect was observed among occasional
heavy drinkers, who consumed 8 or more units per
occasion at least once a month, compared with those
drinkers who did not (OR ¼ 0.89, 95% CI: 0.69–1.15).
Similar risk estimates were generally observed for the
NHL subtypes as for all NHLs combined. Odds ratios
for beverage types were not dissimilar to all alcohol
combined (data not shown). Analyses were repeated for
alcohol consumption by age and sex, at ten and 20 years
preceding diagnosis/reference date, and averaging alco-
hol consumption across the three time periods, where
risk estimates largely remained unchanged (data not
shown).
There was no evidence of interaction between never/
ever smoking status and either frequency (v2 ¼ 3.94,
774 E.V. Willett et al.
Table 1. Distribution of interviewed non-Hodgkin lymphoma cases and controls by confirmed diagnosis, sex, age, marital status, age left school
and deprivation

Variable NHLa

Total DLBCa FCCa MZa Mantle cell T cell NHL nosa Controls

n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%)

Total 700 (100) 318 (100) 228 (100) 73 (100) 28 (100) 35 (100) 18 (100) 915 (100)
Sex
Male 362 (52) 168 (53) 103 (45) 35 (48) 23 (82) 22 (63) 11 (61) 495 (54)
Female 338 (48) 150 (47) 125 (55) 38 (52) 5 (18) 13 (37) 7 (39) 420 (46)
Ageb
18–<35 29 (4) 23 (7) 2 (1) 3 (4) 1 (4) 0 (0) 0 (0) 76 (8)
35–<45 78 (11) 37 (12) 27 (12) 3 (4) 1 (4) 10 (29) 0 (0) 127 (14)
45–<55 244 (35) 108 (34) 93 (41) 23 (32) 6 (21) 9 (26) 5 (28) 297 (32)
55–<65 349 (50) 150 (47) 106 (46) 44 (60) 20 (71) 16 (46) 13 (72) 415 (45)
Marital status
Single 37 (5) 16 (5) 11 (5) 4 (5) 1 (4) 4 (11) 1 (6) 72 (8)
Married/cohabiting 581 (83) 268 (84) 186 (82) 59 (81) 25 (89) 27 (77) 16 (89) 754 (82)
Divorced/separated 59 (8) 19 (6) 25 (11) 9 (12) 2 (7) 3 (9) 1 (6) 68 (7)
Widowed 23 (3) 15 (5) 6 (3) 1 (1) 0 (0) 1 (3) 0 (0) 21 (2)
Age left schoolc
<16 366 (52) 168 (53) 106 (46) 46 (63) 18 (64) 20 (57) 8 (44) 419 (46)
16 198 (28) 95 (30) 68 (30) 19 (26) 5 (18) 7 (20) 4 (22) 292 (32)
>16 136 (19) 55 (17) 54 (24) 8 (11) 5 (18) 8 (23) 6 (33) 204 (22)
Deprivationd
1 (least deprived) 227 (32) 93 (29) 80 (35) 22 (30) 10 (36) 12 (34) 10 (56) 269 (29)
2 156 (22) 77 (24) 46 (20) 18 (25) 5 (18) 9 (26) 1 (6) 232 (25)
3 147 (21) 62 (20) 54 (24) 13 (18) 6 (21) 9 (26) 3 (17) 192 (21)
4 95 (14) 48 (15) 28 (12) 10 (14) 4 (14) 2 (6) 3 (17) 129 (14)
5 (most deprived) 75 (11) 38 (12) 20 (9) 10 (14) 3 (11) 3 (9) 1 (6) 89 (10)
Not classifiable 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 4 (0)

a
NHL: non-Hodgkin lymphoma; DLBC: diffuse large B-cell lymphoma; FCC: follicle centre cell lymphoma; MZ: marginal zone lymphoma;
NHL nos: non-Hodgkin lymphoma not otherwise specified.
b
Age at diagnosis/pseudo diagnosis: cases are more likely to be older than controls (Pearson’s v2 = 15.3, p = 0.002).
c
Age left school: cases are more likely to have left school at younger age than controls (Pearson’s v2 = 6.71, p = 0.035).
d
Deprivation coded using categories of the Townsend scores for England and Wales; cases were no more or less likely to live in deprived areas
than controls (Pearson’s v2 = 6.26, p = 0.28).

p ¼ 0.56), volume (v2 ¼ 5.48, p ¼ 0.24), or pattern
(v2 ¼ 0.47, p ¼ 0.79) of alcohol consumption at 5 years
preceding diagnosis/reference date.
Discussion
This study found little evidence to suggest an association
between NHL and either tobacco smoking or alcohol
consumption. Neither age nor sex modified the risk of
NHL associated with either smoking status or alcohol
consumption. No interaction between smoking status
and alcohol consumption on the risk of NHL was
present. Similar findings were observed for diffuse large
B cell, follicle centre cell, mantle cell, marginal zone and
T-cell lymphomas.
The majority of published epidemiological studies
have reported no association between NHL and tobacco
use [14–21], a few observed increased risks [22–24], one
reported a decreased risk [25], and another reported an
increased risk among men and a decreased risk among
women [26]. Few studies have reported risks for tobacco
consumption stratified by NHL subtypes [21, 22, 27–
29], typically coded to the Working Formulation [30],
and not the WHO classification [7] as here. While diffuse
large B-cell lymphoma and follicle centre cell lymphoma
are not equivalent to specific Working Formulation
subgroups, the majority of diffuse large B-cell and
follicle centre cell lymphomas would have been catego-
rized by the Working Formulation to diffuse lymphoma
and follicular lymphoma respectively [31]. Hence, to
compare risks reported here for diffuse large B-cell and
Table 2. Number of cases and controls, adjusted odds ratios (OR), and 95% confidence intervals (CI) by subtypes of non-Hodgkin lymphoma (NHL) ever smoked tobacco up to 2 years
prior to diagnosis/reference date

Variable Controls NHLa DLBCa FCCa Marginal zone Mantle Cell T cell NHL nosa
(n = 915)
Tobacco and alcohol consumption

Case ORb 95% CI Case ORb 95% CI Case ORb 95% CI Case ORb 95% CI Case ORb 95% CI Case ORb 95% CI Case ORb 95% CI
(n = 700) (n = 318) (n = 228) (n = 73) (n = 28) (n = 35) (n = 18)

Smoking
Never 366 267 1 – 120 1 – 87 1 – 27 1 – 10 1 – 15 1 – 8 1 –
Ever 549 433 1.04 0.85–1.28 198 1.09 0.83–1.42 141 1.09 0.80–1.47 46 1.06 0.64–1.76 18 0.78 0.35–1.77 20 0.84 0.42–1.68 10 0.67 0.25–1.78
Ex 291 225 0.99 0.78–1.26 108 1.11 0.81–1.51 69 0.96 0.67–1.37 23 0.95 0.52–1.73 7 0.50 0.18–1.39 11 0.87 0.38–1.97 7 0.73 0.25–2.12
Current 258 208 1.10 0.86–1.41 90 1.06 0.77–1.46 72 1.24 0.87–1.77 23 1.18 0.65–2.13 11 1.17 0.48–2.88 9 0.80 0.34–1.88 3 0.57 0.14–2.24

Age started smoking

<16 177 146 1.12 0.85–1.47 65 1.12 0.78–1.60 46 1.16 0.77–1.75 15 1.12 0.57–2.21 7 0.87 0.31–2.41 9 1.14 0.48–2.71 4 0.87 0.25–3.12
16–18 234 168 0.96 0.75–1.25 73 0.95 0.68–1.33 56 1.03 0.71–1.51 23 1.30 0.72–2.36 8 0.87 0.33–2.30 5 0.50 0.18–1.40 3 0.50 0.13–1.97
>18 138 119 1.08 0.80–1.46 60 1.29 0.89–1.88 39 1.09 0.71–1.68 8 0.65 0.28–1.47 3 0.53 0.14–2.01 6 1.04 0.39–2.78 3 0.69 0.17–2.75

Years smoked
<10 96 71 1.03 0.73–1.46 37 1.19 0.77–1.84 23 1.02 0.61–1.71 4 0.62 0.21–1.82 2 0.67 0.14–3.17 2 0.51 0.11–2.25 3 1.35 0.34–5.34
10–19 132 94 1.00 0.73–1.37 48 1.13 0.76–1.67 31 1.07 0.68–1.70 7 0.79 0.33–1.86 3 0.69 0.18–2.60 3 0.53 0.15–1.88 2 0.57 0.11–2.81
20–29 111 89 1.10 0.80–1.52 41 1.13 0.74–1.70 29 1.14 0.71–1.84 9 1.10 0.50–2.44 3 0.75 0.20–2.85 5 1.09 0.38–3.07 2 0.74 0.15–3.64
30–39 143 120 1.09 0.81–1.46 51 1.05 0.71–1.56 40 1.17 0.76–1.79 19 1.54 0.82–2.91 4 0.65 0.19–2.15 5 0.86 0.30–2.49 1 0.29 0.03–2.39
‡40 64 59 1.09 0.73–1.64 21 0.94 0.54–1.63 18 1.10 0.61–2.02 7 1.05 0.42–2.63 6 1.33 0.43–4.12 5 1.89 0.59–6.00 2 0.76 0.14–4.03

Years since stopped smoking

‡20 122 86 0.86 0.62–1.19 44 1.07 0.70–1.62 25 0.76 0.46–1.26 8 0.74 0.32–1.72 1 0.15 0.02–1.21 2 0.36 0.08–1.63 6 1.19 0.38–3.72
10–19 91 78 1.11 0.79–1.57 41 1.36 0.88–2.08 23 1.06 0.63–1.78 9 1.19 0.53–2.68 1 0.21 0.03–1.69 3 0.71 0.20–2.57 1 0.37 0.04–3.07
1–9 78 61 1.05 0.73–1.53 23 0.90 0.54–1.50 21 1.18 0.69–2.02 6 1.01 0.40–2.57 5 1.77 0.57–5.53 6 1.82 0.68–4.86 0 0 0–¥
Current 258 208 1.10 0.86–1.41 90 1.06 0.77–1.46 72 1.24 0.87–1.77 23 1.18 0.66–2.14 11 1.17 0.47–2.87 9 0.80 0.34–1.87 3 0.57 0.14–2.22

a
NHL: non-Hodgkin lymphoma; DLBC: diffuse large B-cell lymphoma; FCC: follicle centre cell lymphoma; NHL nos: non-Hodgkin lymphoma not otherwise specified.
b
Odds ratios adjusted for age, sex and region estimated using unconditional logistic regression.
775
 776

Table 3. Number of cases and controls, adjusted odds ratios (OR), and 95% confidence intervals (CI) by subtypes of non-Hodgkin lymphoma (NHL) by frequency, average daily volume
and pattern of heavy drinking of alcohol at 5 years prior to diagnosis/reference date

Variable Control NHLa DLBCa FCCa Marginal zone Mantle cell T Cell NHL nosa
(n = 915)
Case ORb 95% Case ORb 95% Case ORb 95% Case ORb 95% Case ORb 95% Case ORb 95% Case ORb 95%
(n = 700) CI (n = 318) CI (n = 228) CI (n = 73) CI (n = 28) CI (n = 35) CI (n = 18) CI

Frequency
Never ‡43 ‡34 1.03 0.64–1.67 17 1.11 0.61–2.02 12 1.23 0.61–2.47 3 0.63 0.18–2.18 0 0 0–¥ 2 1.28 0.28–5.89 0 0 0–¥
1–5 times a ‡79 ‡84 1.35 0.95–1.93 42 1.49 0.96–2.31 30 1.56 0.95–2.58 7 0.73 0.31–1.75 1 0.61 0.07–5.10 3 1.12 0.30–4.15 1 0.82 0.09–7.61
year
1–2 times a 106 ‡95 1.20 0.87–1.65 39 1.03 0.67–1.57 36 1.58 1.00–2.51 10 0.98 0.46–2.06 5 2.12 0.69–6.52 2 0.52 0.11–2.33 3 2.39 0.53–10.7
month
1–2 times a 340 258 1 – 123 1 – 73 1 – 34 1 – 10 1 – 13 1 – 5 1 –
week
3–4 times a 184 110 0.82 0.62–1.10 49 0.74 0.50–1.08 35 0.98 0.63–1.53 10 0.60 0.29–1.25 3 0.57 0.15–2.14 7 0.95 0.37–2.45 6 3.16 0.91–11.0
week
Most days 163 119 0.94 0.70–1.25 48 0.80 0.54–1.17 42 1.21 0.79–1.86 9 0.53 0.24–1.13 9 1.46 0.57–3.73 8 1.22 0.49–3.03 3 1.16 0.27–5.01

Volume (units per day)

Never ‡43 ‡34 0.91 0.57–1.47 17 0.98 0.54–1.77 12 0.95 0.48–1.88 3 0.79 0.23–2.73 0 0 0–¥ 2 1.26 0.27–5.81 0 0 0–¥
>0–1 359 315 1 – 146 1 – 107 1 – 31 1 – 10 1 – 13 1 – 8 1 –
>1–2 295 198 0.79 0.62–1.01 87 0.72 0.52–0.99 63 0.78 0.55–1.12 25 1.08 0.61–1.92 7 0.64 0.23–1.76 11 0.93 0.40–2.17 5 0.69 0.21–2.27
>2–4 116 ‡85 0.89 0.64–1.25 40 0.84 0.55–1.29 29 1.03 0.63–1.68 6 0.77 0.30–1.97 1 0.24 0.03–1.98 5 1.01 0.34–3.06 4 2.11 0.55–8.04
>4–6 ‡50 ‡33 0.81 0.50–1.31 10 0.48 0.23–1.00 12 1.03 0.51–2.07 5 1.44 0.51–4.11 5 2.78 0.84–9.24 1 0.45 0.06–3.68 0 0 0–¥
>6 ‡52 ‡35 0.84 0.52–1.35 18 0.83 0.46–1.51 5 0.42 0.16–1.11 3 0.80 0.22–2.88 5 2.17 0.65–7.29 3 1.28 0.33–5.00 1 1.36 0.14–12.9

Pattern
Never ‡43 ‡34 0.98 0.61–1.57 17 1.08 0.60–1.94 12 1.00 0.51–1.96 3 0.81 0.24–2.76 0 0 0–¥ 2 1.26 0.29–5.58 0 0 0–¥
<8 units per 616 499 1 – 227 1 – 171 1 – 50 1 – 15 1 – 23 1 – 13 1 –
occasion at
least monthly
‡8 units per 256 167 0.89 0.69–1.15 74 0.79 0.74–1.29 45 0.80 0.54–1.19 20 1.29 0.70–2.37 13 1.96 0.85–4.54 10 0.89 0.39–2.04 5 1.52 0.47–4.91
occasion at
least monthly

a
NHL: non-Hodgkin lymphoma; DLBC: diffuse large B-cell lymphoma; FCC: follicle centre cell lymphoma; NHL nos: non-Hodgkin lymphoma not otherwise specified.
b
Odds ratios, adjusted for age, sex and region estimated using unconditional logistic regression.
E.V. Willett et al.
Tobacco and alcohol consumption
Fig. 1. Study Area.
follicle centre cell lymphomas with those previously
published for diffuse lymphoma and follicular lym-
phoma seems reasonable. Generally, no association with
tobacco smoking has been observed for diffuse lym-
phoma [21, 22, 28, 29], but there has been some sugges-
tion of an increased risk of follicular lymphoma,
although, like our report, these studies largely failed to
observe a convincing dose–response relationship
[21, 27–29]. Other WHO subtypes are poorly defined
under the Working Formulation [32], so to the best of
our knowledge, this is the first study to report risks for
mantle cell, marginal zone and T-cell lymphomas
following tobacco exposure.
777
Compared with persons who did not consume
alcohol regularly, many studies have suggested that
the risk of NHL following consumption of alcohol
may be decreased [18, 33–38], although two studies
reported no association [26, 39]. Where type of alco-
holic beverage was considered, risk estimates mostly
remained below one for wine, but were less consistent
for beer or liquor [18, 26, 34, 36–40]. Risks seldom
differed when reported by the Working Formulation’s
subtypes [33, 34, 36–38]. Unlike previous studies, no
associations were observed here between NHL or its
WHO subtypes and consumption of all or any type of
alcohol. However, in this present study, risks for levels
778
of alcohol consumption at five, ten or 20 years prior
to diagnosis were estimated relative to the commonest
exposure group, which comprised regular consumers
of low levels of alcohol. Previous use of non- or
infrequent drinkers as the referent group may have
underestimated the risk particularly if retrospectively
collected data were not lagged to account for chang-
ing drinking habits as the disease onsets [26, 33, 35–
41].
Little evidence of a dose–response between NHL and
alcohol consumption has been observed [18, 34,
36–39, 41, 42]. Typically, the dose variable was the
average number of drinks or grams of beverage or
ethanol drunk per day, week or month [18, 26, 33–
36, 38, 39]. Such estimates of alcohol intake per unit
time do not account for either the frequency of
consumption or the amount drunk on a given occasion;
for example, a person drinking 2 units of alcohol on
each of seven days would have the same mean volume of
alcohol as a person who drank 14 units of alcohol once
a week. It may be important to differentiate between
light regular drinking and occasional excessive drinking,
since the former may be beneficial to health, while the
latter temporarily affects the normal physiological
function of tissues [8]. Besides the mean volume per
day, frequency and pattern of drinking were also
investigated here; however, no trend was evident with
any dose variable. Of course, a dose–response relation-
ship may be masked if alcohol consumption is under-
reported [43], although, by requesting the customary
alcohol intake by beverage type at certain time-points,
the potential for recall bias may have been reduced [44].
Notification of cases to the present study is high as
employment of the same ascertainment methods have
been estimated to be 96% complete for NHL [45]. Of
more concern is the high number of non-participating
subjects. Cases and controls who refused to participate
tended to live in more deprived areas than those who
were interviewed, and so it was likely that they
smoked more [46] and drank less [47]. As the refusal
rate among controls was considerably greater than
among cases, the resulting risk estimates may have
been biased away from one. Adjustment for depriva-
tion, however, did not alter the OR for either tobacco
(adjusted OR ¼ 1.04, 95% CI: 0.85–1.29) or alcohol
use (current abstainers: adjusted OR ¼ 1.06, 95% CI:
0.65–1.72; one to five times a year: adjusted
OR ¼ 1.36, 95% CI: 0.95–1.95; one to two times a
month: adjusted OR ¼ 1.21, 95% CI: 0.87–1.67; three
to four times a week: adjusted OR ¼ 0.83, 95% CI:
0.61–1.05; most days: adjusted OR ¼ 0.94, 95% CI:
0.69–1.21). Another limitation of this study is the
retrospective collection of self-reported exposure infor-
E.V. Willett et al.
mation. Since the hypotheses that either smoking or
alcohol is related to NHL are not widely known,
exaggeration of tobacco and alcohol use by cases
seems unlikely. Controls, on the other hand, may have
under-reported their consumption of tobacco and
alcohol. However, controls here reported no less
exposure to tobacco and alcohol than was reported
by a sample of the British population [46, 47].
In conclusion, we found no evidence to suggest that
NHL was associated with tobacco smoking. There was
no suggestion of a trend with age started smoking, the
number of years smoked, the number of years since
smoking ceased, or the number of cigarettes smoked
per day. Similarly, no associations were observed
between NHL and moderate consumption of alcohol,
despite considering the frequency of alcohol consump-
tion and the pattern of heavy drinking, as well as the
average number of drinks per day. Indeed, this study
is consistent with the view that NHL is unlikely to be
associated with tobacco smoking [48] or alcohol [49].
Nevertheless, it remains that, for these factors, the
heterogeneous NHL subtypes have been poorly stud-
ied. Only through collaborative study such as Inter-
Lymph [50], to which the reported data will be added,
can greater power to investigate the risk of specific
WHO subtypes be achieved.
Acknowledgements
Authors thank all consultants, hospital staff, general
practitioners, and interviewees who participated in the
study. Our thanks also go to the study staff: A Preston,
B Routledge, S Griffiths, J O’Sullivan, S Pope, P Sanders,
P Johnson, S Muir, B Longshaw, I Cope, B Cooper,
D Robinson, V Sinclair, D White, A McKeating, CI
McAlpine, L Kimpton and J Prajapati. We would also
like to acknowledge Andrew Jack and Bridget Wilkins
for confirming the patients’ diagnoses and Ray Cart-
wright for his assistance in the design of the study.
References
1. Hersey P, Prendergast D, Edwards A (1983) Effects of cigarette
smoking on the immune system. Follow-up studies in normal
subjects after cessation of smoking. Med J Aust 2: 425–429.
2. Mufti SI, Prabhala R, Moriguchi S, Sipes IG, Watson RR (1988)
Functional and numerical alterations induced by ethanol in the
cellular immune system. Immunopharmacology 15: 85–93.
3. Percival SS, Sims CA (2000) Wine modifies the effects of alcohol
on immune cells of mice. J Nutr 130: 1091–1094.
4. Meliska CJ, Stunkard ME, Gilbert DG, Jensen RA, Martinko JM
(1995) Immune function in cigarette smokers who quit smoking for
31 days. J Allergy Clin Immunol 95: 901–910.
Tobacco and alcohol consumption
5. McAllister-Sistilli CG, Caggiula AR, Knopf S, Rose CA, Miller AL,
Donny EC (1998) The effects of nicotine on the immune system.
Psychoneuroendocrinology 23: 175–187.
6. Richards SJ, Jack AS (2003) The development of integrated
haematopathology laboratories: a new approach to the diagnosis
of leukaemia and lymphoma. Clin Lab Haematol 25: 337–342.
7. Fritz A, Percy C, Jack A, et al. (2000) International Classification
for Diseases for Oncology. Geneva: World Health Organisation.
8. Inter-departmental Working Group (1995). Sensible Drinking.
London: Department of Health.
9. Rehm J, Greenfield TK, Rogers JD (2001) Average volume of
alcohol consumption, patterns of drinking, and all-cause mortality:
results from the US National Alcohol Survey. Am J Epidemiol 153:
64–71.
10. Shaper AG (1995) The ‘‘unhealthy abstainers’’ question is still
important. Addiction 90: 488–490.
11. Townsend P, Phillimore P, Beattie A (1988) Health and Depriva-
tion: Inequality and the North. London: Croom Helm.
12. Breslow NE, Day NE (1980) Statistical Methods in Cancer
Research. Vol. 1–The Analysis of Case-Control Studies. Lyon:
International Agency for Research on Cancer.
13. StataCorp (2003). Stata Statistical Software: Release 8.2. College
Station, Texas: Stata Corporation.
14. Paffenbarger RS Jr, Wing AL, Hyde RT (1978) Characteristics
in youth predictive of adult-onset malignant lymphomas, mel-
anomas, and leukemias: brief communication. J Natl Cancer Inst
60: 89–92.
15. Tavani A, Negri E, Franceschi S, Serraino D, La Vecchia C (1994)
Smoking habits and non-Hodgkin’s lymphoma: a case-control
study in northern Italy. Prev Med 23: 447–452.
16. McLaughlin JK, Hrubec Z, Blot WJ, Fraumeni JF Jr (1995)
Smoking and cancer mortality among US veterans: a 26-year
follow-up. Int J Cancer 60: 190–193.
17. Siemiatycki J, Krewski D, Franco E, Kaiserman M (1995)
Associations between cigarette smoking and each of 21 types of
cancer: a multi-site case-control study. Int J Epidemiol 24: 504–514.
18. Nelson RA, Levine AM, Marks G, Bernstein L (1997) Alcohol,
tobacco and recreational drug use and the risk of non- Hodgkin’s
lymphoma. Br J Cancer 76: 1532–1537.
19. Freedman DS, Tolbert PE, Coates R, Brann EA, Kjeldsberg CR
(1998) Relation of cigarette smoking to non-Hodgkin’s lymphoma
among middle-aged men. Am J Epidemiol 148: 833–841.
20. Fabbro-Peray P, Daures JP, Rossi JF (2001) Environmental risk
factors for non-Hodgkin’s lymphoma: a population- based case-
control study in Languedoc-Roussillon, France. Cancer Causes
Control 12: 201–212.
21. Morton LM, Holford TR, Leaderer B, et al. (2003) Cigarette
smoking and risk of non-Hodgkin lymphoma subtypes among
women. Br J Cancer 89: 2087–2092.
22. Brown LM, Everett GD, Gibson R, Burmeister LF, Schuman LM,
Blair A (1992) Smoking and risk of non-Hodgkin’s lymphoma and
multiple myeloma. Cancer Causes Control 3: 49–55.
23. Linet MS, McLaughlin JK, Hsing AW, et al. (1992) Is cigarette
smoking a risk factor for non-Hodgkin’s lymphoma or multiple
myeloma? Results from the Lutheran Brotherhood Cohort Study.
Leuk Res 16: 621–624.
24. Miligi L, Seniori CA, Crosignani P, et al. (1999) Occupational,
environmental, and life-style factors associated with the risk of
hematolymphopoietic malignancies in women. Am J Ind Med 36:
60–69.
25. Cartwright RA, McKinney PA, O’Brien C, et al. (1988) Non-
Hodgkin’s lymphoma: case control epidemiological study in
Yorkshire. Leuk Res 12: 81–88.
779
26. De Stefani E, Fierro L, Barrios E, Ronco A (1998) Tobacco,
alcohol, diet and risk of non-Hodgkin’s lymphoma: a case-control
study in Uruguay. Leuk Res 22: 445–452.
27. Herrinton LJ, Friedman GD (1998) Cigarette smoking and risk of
non-Hodgkin’s lymphoma subtypes. Cancer Epidemiol Biomarkers
Prev 7: 25–28.
28. Parker AS, Cerhan JR, Dick F, et al. (2000) Smoking and risk of
non-Hodgkin lymphoma subtypes in a cohort of older women.
Leuk Lymph 37: 341–349.
29. Stagnaro E, Ramazzotti V, Crosignani P, et al. (2001) Smoking
and hematolymphopoietic malignancies. Cancer Causes Control
12: 325–334.
30. Anonymous (1982) National Cancer Institute sponsored study of
classifications of non-Hodgkin’s lymphomas: summary and descrip-
tion of a working formulation for clinical usage. The Non-Hodgkin’s
Lymphoma Pathologic Classification Project. Cancer 49: 2112–2135.
31. Pittaluga S, Bijnens L, Teodorovic I, et al. (1996) Clinical analysis
of 670 cases in two trials of the European Organization for the
Research and Treatment of Cancer Lymphoma Cooperative
Group subtyped according to the Revised European-American
Classification of Lymphoid Neoplasms: a comparison with the
Working Formulation. Blood 87: 4358–4367.
32. Herrinton LJ (1998) Epidemiology of the Revised European-
American Lymphoma Classification subtypes. Epidemiol Rev 20:
187–203.
33. Brown LM, Gibson R, Burmeister LF, Schuman LM, Everett GD,
Blair A (1992) Alcohol consumption and risk of leukemia, non-
Hodgkin’s lymphoma, and multiple myeloma. Leuk Res 16: 979–984.
34. Chiu BC, Cerhan JR, Gapstur SM, et al. (1999) Alcohol con-
sumption and non-Hodgkin lymphoma in a cohort of older
women. Br J Cancer 80: 1476–1482.
35. Holly EA, Lele C, Bracci PM, McGrath MS (1999) Case-control
study of non-Hodgkin’s lymphoma among women and heterosex-
ual men in the San Francisco Bay Area, California. Am J
Epidemiol 150: 375–389.
36. Chiu BC, Weisenburger DD, Cantor KP, et al. (2002) Alcohol
consumption, family history of hematolymphoproliferative cancer,
and the risk of non-Hodgkin’s lymphoma in men. Ann Epidemiol
12: 309–315.
37. Briggs NC, Levine RS, Bobo LD, Haliburton WP, Brann EA,
Hennekens CH (2002) Wine drinking and risk of non-Hodgkin’s
lymphoma among men in the United States: a population-based
case-control study. Am J Epidemiol 156: 454–462.
38. Morton LM, Holford TR, Leaderer B, et al. (2003) Alcohol use
and risk of non-Hodgkin’s lymphoma among Connecticut women
(United States). Cancer Causes Control 14: 687–694.
39. Tavani A, Gallus S, La Vecchia C, Franceschi S (2001) Alcohol
drinking and risk of non-Hodgkin’s lymphoma. Eur J Clin Nutr 55:
824–826.
40. Franceschi S, Serraino D, Carbone A, Talamini R, La Vecchia C
(1989) Dietary factors and non-Hodgkin’s lymphoma: a case-
control study in the northeastern part of Italy. Nutr Cancer 12:
333–341.
41. Tavani A, Pregnolato A, Negri E et al. (1997) Diet and risk of
lymphoid neoplasms and soft tissue sarcomas. Nutr Cancer 27:
256–260.
42. Franceschi S, Serraino D, Bidoli E, et al. (1989) The epidemiology
of non-Hodgkin’s lymphoma in the north-east of Italy: a hospital-
based case-control study. Leuk Res 13: 465–472.
43. Rothman KJ, Greenland S (1998) Modern Epidemiology. Lippin-
cott-Raven Publishers.
44. Dawson DA (1998) Measuring alcohol consumption: limitations
and prospects for improvement. Addiction 93: 965–968.
780
45. McNally RJ, Rowland D, Roman E, Cartwright RA (1997) Age
and sex distributions of hematological malignancies in the UK.
Hematol Oncol 15: 173–189.
46. Walker A, O’Brien M, Traynor J, Fox K, Goddard E, Foster K
(2002) Smoking. Living in Britain: Results from the 2001 General
Household Survey. London: HMSO, pp. 115–138.
47. Walker A, O’Brien M, Traynor J, Fox K, Goddard E, Foster K
(2002) Drinking. Living in Britain: Results from the 2001 General
Household Survey. London: HMSO, pp. 139–162.
E.V. Willett et al.
48. Kuper H, Boffetta P, Adami HO (2002) Tobacco use and cancer
causation: association by tumour type. J Intern Med 252: 206–
224.
49. Ringborg U (1998) Alcohol and risk of cancer. Alcohol Clin Exp
Res 22: 323S–328S.
50. Boffetta P, Linet M, Armstrong B (2003). The InterLymph
collaboration: a consortium of molecular epidemiological studies
of non-Hodgkin’s lymphoma. Proc Am Assoc Cancer Res
44:1579.