Abstract

Cigar and pipe smoking are considered risk factors for head and neck cancers, but the magnitude of effect estimates for these products has been imprecisely estimated. By using pooled data from the International Head and Neck Cancer Epidemiology (INHANCE) Consortium (comprising 13,935 cases and 18,691 controls in 19 studies from 1981 to 2007), we applied hierarchical logistic regression to more precisely estimate odds ratios and 95% confidence intervals for cigarette, cigar, and pipe smoking separately, compared with reference groups of those who had never smoked each single product. Odds ratios for cigar and pipe smoking were stratified by ever cigarette smoking. We also considered effect estimates of smoking a single product exclusively versus never having smoked any product (reference group). Among never cigarette smokers, the odds ratio for ever cigar smoking was 2.54 (95% confidence interval (CI): 1.93, 3.34), and the odds ratio for ever pipe smoking was 2.08 (95% CI: 1.55, 2.81). These odds ratios increased with increasing frequency and duration of smoking (Ptrend ≤ 0.0001). Odds ratios for cigar and pipe smoking were not elevated among ever cigarette smokers. Head and neck cancer risk was elevated for those who reported exclusive cigar smoking (odds ratio = 3.49, 95% CI: 2.58, 4.73) or exclusive pipe smoking (odds ratio = 3.71, 95% CI: 2.59, 5.33). These results suggest that cigar and pipe smoking are independently associated with increased risk of head and neck cancers.

Head and neck cancers are among the most common cancers worldwide with nearly 600,000 new cases and 300,000 deaths occurring globally each year (1, 2). These cancers include cancers of the oral cavity, pharynx, and larynx (2). Approximately 90% of head and neck cancers are squamous cell carcinomas (2). Associations between cigarette smoking and the incidence and mortality of these cancers are well established (3). Although cigar and pipe smoking are not as common as cigarette smoking, previous research has suggested that cigar and pipe smoke may contain equivalent, or in some instances higher, doses of carcinogens such as benzo[a]pyrene compared with cigarette smoke (3). Epidemiologic evidence supports associations between cigar or pipe smoking and the risk of head and neck cancers; however, previous estimates were frequently based on studies with small sample sizes and sometimes limited information on important covariates (e.g., frequency and duration of cigarette smoking and alcohol drinking) (3). Furthermore, it has been difficult to focus on individuals who smoke cigars or pipes exclusively (3). Therefore, we used pooled data from the International Head and Neck Cancer Epidemiology (INHANCE) Consortium to better characterize the individual and combined associations of cigarette, cigar, and pipe smoking with the risk of head and neck cancers.

MATERIALS AND METHODS

Study design

By using pooled data (data version 1.2) from the INHANCE Consortium, we analyzed the risk of head and neck cancers from smoking cigarettes, cigars, or pipes among 13,935 cases and 18,691 controls from 13 case-control studies conducted in Milan, Italy (1984–1989); Aviano, Italy (1987–1992); Vaud, Switzerland (1991–1997); Seattle, Washington (1985–1995); Iowa (1993–2006); North Carolina (1994–1997); Tampa, Florida (1994–2003); Los Angeles, California (1999–2004); Houston, Texas (2001–2006); Puerto Rico (1992–1995); Rome, Italy (2002–2007); New York, New York (1992–1994); and São Paulo, Brazil (2002–2007) and also included 6 multicenter case-control studies from Italy (1990–1999), Central Europe (1998–2003), New York State (1981–1990), Latin America (2000–2003), and the United States (1983–1984) and an international, multicountry study (1992–1997) (4). For the present analysis, we excluded 1 study from France that participates in the INHANCE Consortium because only smokers were enrolled (5), as well as 1 study from Boston, Massachusetts, because it had not been published at the commencement of this analysis. Individuals from the Sudanese (106 cases and 151 controls) and Indian (576 cases and 582 controls) centers from the International Agency for Research on Cancer's (Lyon, France) international multicountry study were also excluded, because the tobacco products used in these areas differ from those of other regions of the world, and an analysis of the tobacco data from Indian centers was previously published (3, 6).

The individual studies and a complete description of the pooling methods used in the INHANCE Consortium have been previously described (4, 7). Briefly, most studies were hospital based with controls that were frequency matched to cases on the basis of age and sex. Although the exact wording of tobacco-related questions differed across studies, questions were conceptually similar. For example, each study asked about ever having smoked cigarettes, cigars, or pipes, but definitions of ever smoking varied among the individual studies. The exact definitions of ever smoking each product for individual studies are detailed in previous INHANCE Consortium publications (4, 7). Variables on the frequency (number of cigarettes, cigars, or pipes smoked per day), duration (in years), and cumulative duration (in pack-years, cigar-years, or pipe-years) of tobacco smoking were also ascertained in each study. Pack-years were calculated as packs of cigarettes smoked per day multiplied by the years of cigarette smoking, where 20 cigarettes equaled 1 pack. Cigar-years and pipe-years were calculated by multiplying the frequency and duration of smoking that specific product. Levels of frequency, duration, and cumulative use of tobacco were categorized according to previous INHANCE Consortium reports (7) except where sparse strata required collapsing of levels, such as in the frequency of cigar and pipe smoking.

Cases of invasive head and neck cancers were categorized by tumor site according to the International Classification of Diseases for Oncology, Second Edition, or the International Classification of Diseases, Ninth or Tenth Edition (7–10). We included cancers of the oral cavity (codes C00.3–C00.9, C02.0–C02.3, C03.0, C03.1, C03.9, C04.0, C04.1, C04.8, C04.9, C05.0, C06.0–C06.2, C06.8, and C06.9), oropharynx (codes C01.9, C02.4, C05.1, C05.2, C09.0, C09.1, C09.8, C09.9, C10.0–C10.4, C10.8, and C10.9), hypopharynx (codes C12.9, C13.0–C13.2, C13.8, and C13.9), oral cavity or pharynx overlapping or not otherwise specified (codes C02.8, C02.9, C05.8, C05.9, C14.0, C14.2, and C14.8), larynx (codes C32.0–C32.3 and C32.8–C32.9), or head and neck cancers that were overlapping or unspecified (7). Cancers of the salivary glands, lip, nasopharynx, and esophagus were excluded (7). Head and neck cancers, regardless of site and histology, were the primary outcome of interest. However, analyses with site-specific outcomes (oral cavity, pharynx, oropharynx, hypopharynx, larynx, or not otherwise specified) and restricted to squamous cell carcinomas of the head and neck were also considered.

Individuals with missing data on sex, age, race/ethnicity, or subtype of cancer (140 cases and 82 controls) were excluded from all analyses; because this was a small proportion of individuals in this pooled analysis (1.0% of cases and 0.4% of controls), these exclusions did not likely materially affect the results. Subjects from the Central European and Latin American multicenter studies were classified as non-Hispanic white and Latin American, respectively, because information on race and ethnicity was not collected in these studies. Information on educational level was not collected in the Rome, Italy, study and was frequently missing in the Latin American multicenter study. For study participants with missing information on educational level, we used single imputation (MI procedure) in SAS, version 9.2, software (SAS Institute, Inc., Cary, North Carolina) to predict educational level within geographical regions on the basis of age, sex, race/ethnicity, study center, and case or control status (7).

Statistical analysis

We estimated odds ratios and 95% confidence intervals for ever tobacco users and levels of the frequency, duration, and cumulative use for each tobacco product compared with never users of that same product by using hierarchical logistic regression with study center as a random effect. Levels of frequency, duration, and cumulative tobacco use were modeled by using disjoint indicator variables except for linear trends, which were assessed through P values obtained from modeling the continuous forms of those variables. We incorporated a random intercept for study center into the model to account for clustering of data by study center and because we detected significant heterogeneity across studies. Because we included a random intercept for study center, we excluded from the analyses of cigar smoking 811 cases and 1,095 controls from centers in the Latin American, Central European, and international multicenter studies that did not report any cases or any controls who smoked cigars (study centers in Pelotas, Brazil; Bucharest, Romania; Budapest, Hungary; Lodz, Poland; Banska Bystrica, Slovakia; Aviano, Italy; Udine, Italy; Milan, Italy; Poland; and Australia). Likewise, we excluded 625 cases and 735 controls from the analyses of pipe smoking from centers that did not report any cases or controls who smoked pipes (Pelotas, Brazil; Budapest, Hungary; Lodz, Poland; Udine, Italy; Granada, Spain; and Poland). These exclusions did not materially change the magnitude of the odds ratios. All analyses were performed in SAS software (SAS Institute, Inc.) by using the NLMIXED procedure (initial parameter values for the NLMIXED procedure models were based on parameter estimates from logistic procedure models, and an initial variance value of 2.0 was used; varying this starting point did not sustainably change final estimates) (11).

Odds ratios were adjusted for sex, age (continuous years), race (categorical indicator for white, black, Hispanic, Asian or Pacific Islander, Latin American, or other), educational level (categorical indicator for no education or junior high school or less, some high school, high school graduate, technical school or some college, or college graduate or more), and frequency of alcohol use (mL of ethanol per day truncated at the 95th percentile among alcohol drinkers to account for extreme values) (7). Because the reference category for initial analyses was never smokers of only the product of interest, cigarette smoking odds ratios were also adjusted for the duration of cigar smoking and the duration of pipe smoking (both in years); cigar smoking odds ratios were also adjusted for the duration of pipe smoking and the duration of cigarette smoking (both in years); and pipe smoking odds ratios were also adjusted for the duration cigar smoking and the duration of cigarette smoking (both in years). Additionally, cigar and pipe analyses were stratified by ever cigarette smoking to further control for the confounding effect of cigarette smoking (duration of cigarette smoking was removed from the adjustment set for odds ratios among never cigarette smokers). Cigarette, cigar, and pipe smoking odds ratios were not adjusted for smokeless tobacco use because information on smokeless tobacco use was available only for US studies; adjustment for smokeless tobacco use in the analyses restricted to US studies did not substantially change estimates for smoking tobacco products. Likewise, models were not adjusted for marijuana use because few studies collected information on this variable (12). We also considered analyses stratified by region (Europe, North America, or Latin and South America), sex (male or female with sex removed from the adjustment set), and age (<45 years or ≥45 years with continuous age removed from the adjustment set).

Joint and exclusive smoking of tobacco products was also modeled. For this set of analyses, joint and exclusive smoking of products was defined as ever having smoked the specified tobacco product(s) and never having smoked all other tobacco products. The reference group comprised never smokers of any product. For example, smoking of only cigars was defined as ever having smoked cigars and never having smoked cigarettes or pipes with a reference group of never smokers of cigarettes, cigars, or pipes. In addition, we modeled dominant and predominant lifetime consumption of individual smoking products. To standardize consumption across products, we calculated product-specific lifetime consumption (in grams of tobacco) on the basis of International Agency for Research on Cancer estimates of average unit weights as follows (3): frequency of cigarette smoking (number of cigarettes per day) × duration of cigarette smoking (years) × 1 (assuming the average weight of tobacco in a cigarette was 1 g); frequency of cigar smoking (number of cigars per day) × duration of cigar smoking (years) × 4 (assuming the average weight of tobacco in a cigar was 4 g); and frequency of pipe smoking (number of pipes per day) × duration of pipe smoking (years) × 3.5 (assuming the average weight of tobacco in a pipe fill was 3.5 g). Total lifetime consumption of smoking tobacco equaled the sum of product-specific lifetime consumptions for cigarettes, cigars, and pipes. Product-specific lifetime consumptions of cigarettes, cigars, and pipes were then divided by the total lifetime consumption of smoking products. Dominant smoking behaviors for each product were defined as 100% lifetime consumption of only that 1 product. For example, if lifetime consumption of cigars divided by total lifetime consumption of smoking tobacco equaled 1, then the tobacco user was a dominant cigar smoker. Predominant smoking behaviors for each product were defined as 66.6%–99.9% of lifetime consumption of only that 1 product. For example, if lifetime consumption of cigars divided by total lifetime consumption of smoking tobacco equaled 0.66–0.99, then the tobacco user was a predominant cigar smoker. The reference group for dominant/predominant smokers was never smokers, who were those with lifetime consumption of 0 g of cigarettes, cigars, and pipes. Odds ratios for exclusive/joint and dominant/predominant use were adjusted for sex, age, race, education, and frequency of alcohol use as previously described.

Institutional review board approval was obtained at each study center, and all identifying information was removed before data were received for pooling. In addition, institutional review board approval was obtained for this analysis from the University of North Carolina at Chapel Hill (Chapel Hill, North Carolina).

RESULTS

Characteristics of cases and controls in this pooled analysis are presented in Table 1. Briefly, one-third of all subjects were from the United States (35% of cases and 36% of controls), and most of the remaining subjects were from Latin America (28% of cases and 17% of controls) or Italy (18% of cases and 29% of controls). Approximately half of the cases (50.6%) and controls (46.7%) were between the ages of 50 and 65 years, and the majority were non-Hispanic white (64.7% of cases and 75.6% of controls). Consistent with the literature, there were more male cases (79.7%) than female cases (20.3%) (2, 13).

Table 1.

Characteristics of Head and Neck Cancer Cases and Controls in the International Head and Neck Cancer Epidemiology Consortium, 1981–2007a

Characteristic No. of Cases No. of Controls 
Total 13,935  18,691  
Study location     
 Aviano, Italy 482 3.5 855 4.6 
 Central Europe (multicenter) 762 5.5 907 4.9 
 Houston, Texas 829 5.9 865 4.6 
 IARC (international multicenter) 881 6.3 992 5.3 
 Iowa 546 3.9 759 4.1 
 Italy (multicenter) 1,261 9.0 2,716 14.5 
 Latin America (multicenter) 2,191 15.7 1,706 9.1 
 Los Angeles, California 417 3.0 1,004 5.4 
 Milan, Italy 416 3.0 1,531 8.2 
 New York State (multicenter) 1,118 8.0 906 4.8 
 New York, New York (MSKCC) 133 1.0 168 0.9 
 North Carolina 180 1.3 202 1.1 
 Puerto Rico 350 2.5 521 2.8 
 Rome, Italy 361 2.6 396 2.1 
 São Paulo, Brazil 1,764 12.7 1,508 8.1 
 Seattle, Washington 407 2.9 607 3.2 
 Switzerland 516 3.7 883 4.7 
 Tampa, Florida 207 1.5 897 4.8 
 United States (multicenter) 1,114 8.0 1,268 6.8 
Race/ethnicityb     
 Asian/Pacific Islander 49 0.4 84 0.4 
 Black 581 4.2 674 3.6 
 Hispanic/Latino 213 1.5 429 2.3 
 Latin American 3,955 28.4 3,214 17.2 
 Non-Hispanic white 9,014 64.7 14,124 75.6 
 Other 123 0.9 166 0.9 
Age, years     
 17–39 502 3.6 1,356 7.3 
 40–44 766 5.5 1,345 7.2 
 45–49 1,508 10.8 2,042 10.9 
 50–54 2,169 15.6 2,808 15.0 
 55–59 2,537 18.2 3,108 16.6 
 60–64 2,340 16.8 2,827 15.1 
 65–69 1,936 13.9 2,408 12.9 
 70–74 1,341 9.6 1,774 9.5 
 75–96 836 6.0 1,023 5.5 
Sex     
 Male 11,108 79.7 13,025 69.7 
 Female 2,827 20.3 5,666 30.3 
Educational levelc     
 No formal education 280 2.0 190 1.0 
 Less than junior high school 5,656 40.6 7,081 37.9 
 Some high school 2,077 14.9 2,267 12.1 
 High school graduate 1,849 13.3 2,280 12.2 
 Vocational school, some college 2,068 14.8 3,456 18.5 
 College graduate/postgraduate 1,278 9.2 2,768 14.8 
 Missing 730 5.2 651 3.5 
Tumor site     
 Head and neck (not otherwise specified) 314 2.3   
 Hypopharynx 1,026 7.4   
 Larynx 3,461 24.8   
 Oral cavity 4,110 29.5   
 Oral/pharynx (not otherwise specified) 1,190 8.5   
 Oropharynx 3,834 27.5   
Characteristic No. of Cases No. of Controls 
Total 13,935  18,691  
Study location     
 Aviano, Italy 482 3.5 855 4.6 
 Central Europe (multicenter) 762 5.5 907 4.9 
 Houston, Texas 829 5.9 865 4.6 
 IARC (international multicenter) 881 6.3 992 5.3 
 Iowa 546 3.9 759 4.1 
 Italy (multicenter) 1,261 9.0 2,716 14.5 
 Latin America (multicenter) 2,191 15.7 1,706 9.1 
 Los Angeles, California 417 3.0 1,004 5.4 
 Milan, Italy 416 3.0 1,531 8.2 
 New York State (multicenter) 1,118 8.0 906 4.8 
 New York, New York (MSKCC) 133 1.0 168 0.9 
 North Carolina 180 1.3 202 1.1 
 Puerto Rico 350 2.5 521 2.8 
 Rome, Italy 361 2.6 396 2.1 
 São Paulo, Brazil 1,764 12.7 1,508 8.1 
 Seattle, Washington 407 2.9 607 3.2 
 Switzerland 516 3.7 883 4.7 
 Tampa, Florida 207 1.5 897 4.8 
 United States (multicenter) 1,114 8.0 1,268 6.8 
Race/ethnicityb     
 Asian/Pacific Islander 49 0.4 84 0.4 
 Black 581 4.2 674 3.6 
 Hispanic/Latino 213 1.5 429 2.3 
 Latin American 3,955 28.4 3,214 17.2 
 Non-Hispanic white 9,014 64.7 14,124 75.6 
 Other 123 0.9 166 0.9 
Age, years     
 17–39 502 3.6 1,356 7.3 
 40–44 766 5.5 1,345 7.2 
 45–49 1,508 10.8 2,042 10.9 
 50–54 2,169 15.6 2,808 15.0 
 55–59 2,537 18.2 3,108 16.6 
 60–64 2,340 16.8 2,827 15.1 
 65–69 1,936 13.9 2,408 12.9 
 70–74 1,341 9.6 1,774 9.5 
 75–96 836 6.0 1,023 5.5 
Sex     
 Male 11,108 79.7 13,025 69.7 
 Female 2,827 20.3 5,666 30.3 
Educational levelc     
 No formal education 280 2.0 190 1.0 
 Less than junior high school 5,656 40.6 7,081 37.9 
 Some high school 2,077 14.9 2,267 12.1 
 High school graduate 1,849 13.3 2,280 12.2 
 Vocational school, some college 2,068 14.8 3,456 18.5 
 College graduate/postgraduate 1,278 9.2 2,768 14.8 
 Missing 730 5.2 651 3.5 
Tumor site     
 Head and neck (not otherwise specified) 314 2.3   
 Hypopharynx 1,026 7.4   
 Larynx 3,461 24.8   
 Oral cavity 4,110 29.5   
 Oral/pharynx (not otherwise specified) 1,190 8.5   
 Oropharynx 3,834 27.5   

Abbreviations: IARC, International Agency for Research on Cancer; MSKCC, Memorial Sloan-Kettering Cancer Center.

a All centers in the International Head and Neck Cancer Epidemiology Consortium studies (data version 1.2) were included except those in Boston, Massachusetts; France; Sudan; and India. The French center was not included because it enrolled only smokers. The Boston, Massachusetts, center was not included because its data are currently unpublished. The centers in Sudan and India from the International Agency for Research on Cancer's international multicenter study were not included because of differences in tobacco products in these regions.

b Information on race and ethnicity was not collected in the Central European and Latin American multicenter studies. All cases and controls were classified as non-Hispanic white for the Central European study and as Latin American for the Latin American study.

c Educational information in this table represents actual counts and percentages; however, because educational information was not collected in the Rome, Italy, study and was frequently missing in the Latin American study, we imputed educational levels during subsequent analyses. Three cases and 2 controls who were originally missing information on educational level were also missing imputed educational level and were therefore excluded from subsequent analyses.

The overall odds ratio of head and neck cancers was elevated among ever cigarette smokers compared with never cigarette smokers, even after adjustment for smoking of other products (odds ratio (OR) = 3.46, 95% confidence interval (CI): 3.24, 3.70) (Table 2). We also observed increasing risk with increasing frequency, duration, and pack-years of cigarette smoking (Ptrend < 0.0001) (Table 2). After stratification by ever cigarette smoking, the adjusted odds ratios for cigar and pipe smokers were increased among never cigarette smokers but not among ever cigarette smokers (Tables 3 and 4). Among never cigarette smokers, the adjusted odds ratios for ever cigar and ever pipe smokers were 2.54 (95% CI: 1.93, 3.34) and 2.08 (95% CI: 1.55, 2.81), respectively. Among ever cigarette smokers, the adjusted odds ratios for ever cigar and ever pipe smokers were 0.90 (95% CI: 0.78, 1.02) and 0.81 (95% CI: 0.70, 0.93), respectively. With respect to frequency, duration, and cumulative use, head and neck cancer risk increased with increasing cigar smoking and pipe smoking among never cigarette smokers (Ptrend ≤ 0.0001), but we did not observe comparable dose-response trends among ever cigarette smokers (Tables 3 and 4).

Table 2.

Odd Ratios for Cigarette Smoking and Head and Neck Cancer in the International Head and Neck Cancer Epidemiology Consortium, 1981–2007

Variable Cigarette Smoking
 
No. of Cases No. of Controls ORa 95% CI 
Tobacco smokingb     
 Never 1,817 7,387 1.00 Referent 
 Ever 12,052 11,277 3.46 3.24, 3.70 
 Missing 66 27   
Smoking frequency (cigarettes per day)     
 Never 1,817 7,387 1.00 Referent 
 1–10 1,620 3,158 1.88 1.72, 2.05 
 11–20 5,127 4,721 3.87 3.59, 4.17 
 21–30 2,148 1,373 5.30 4.81, 5.85 
 31–40 2,020 1,173 5.37 4.85, 5.95 
 >40 996 711 4.02 3.54, 4.58 
 Missing 207 168   
  Ptrendc   <0.0001 
Smoking duration, years     
 Never 1,817 7,387 1.00 Referent 
 1–10 382 1,346 1.04 0.90, 1.19 
 11–20 721 2,021 1.37 1.23, 1.53 
 21–30 2,258 2,721 2.95 2.71, 3.22 
 31–40 4,048 2,751 5.02 4.62, 5.45 
 >40 4,528 2,338 6.55 6.01, 7.15 
 Missing 181 127   
  Ptrend   <0.0001 
Cumulative smoking, pack-years     
 Never 1,817 7,387 1.00 Referent 
 1–10 838 2,525 1.29 1.16, 1.43 
 11–20 1,213 2,120 2.20 1.99, 2.42 
 21–30 1,779 1,890 3.53 3.21, 3.88 
 31–40 2,085 1,538 4.93 4.48, 5.42 
 41–50 1,743 1,025 5.85 5.26, 6.50 
 >50 4,205 1,986 6.81 6.24, 7.44 
 Missing 255 220   
  Ptrend   <0.0001 
Variable Cigarette Smoking
 
No. of Cases No. of Controls ORa 95% CI 
Tobacco smokingb     
 Never 1,817 7,387 1.00 Referent 
 Ever 12,052 11,277 3.46 3.24, 3.70 
 Missing 66 27   
Smoking frequency (cigarettes per day)     
 Never 1,817 7,387 1.00 Referent 
 1–10 1,620 3,158 1.88 1.72, 2.05 
 11–20 5,127 4,721 3.87 3.59, 4.17 
 21–30 2,148 1,373 5.30 4.81, 5.85 
 31–40 2,020 1,173 5.37 4.85, 5.95 
 >40 996 711 4.02 3.54, 4.58 
 Missing 207 168   
  Ptrendc   <0.0001 
Smoking duration, years     
 Never 1,817 7,387 1.00 Referent 
 1–10 382 1,346 1.04 0.90, 1.19 
 11–20 721 2,021 1.37 1.23, 1.53 
 21–30 2,258 2,721 2.95 2.71, 3.22 
 31–40 4,048 2,751 5.02 4.62, 5.45 
 >40 4,528 2,338 6.55 6.01, 7.15 
 Missing 181 127   
  Ptrend   <0.0001 
Cumulative smoking, pack-years     
 Never 1,817 7,387 1.00 Referent 
 1–10 838 2,525 1.29 1.16, 1.43 
 11–20 1,213 2,120 2.20 1.99, 2.42 
 21–30 1,779 1,890 3.53 3.21, 3.88 
 31–40 2,085 1,538 4.93 4.48, 5.42 
 41–50 1,743 1,025 5.85 5.26, 6.50 
 >50 4,205 1,986 6.81 6.24, 7.44 
 Missing 255 220   
  Ptrend   <0.0001 

Abbreviations: CI, confidence interval; OR, odds ratio.

a Cigarette smoking odds ratios are adjusted for sex, age, race, educational level, study center, frequency of alcohol use (mL of ethanol per day), duration of cigar smoking (in years), and duration of pipe smoking (in years). Individuals missing the main variable or 1 or more covariates were dropped from the models. The most frequently missing variable was frequency of alcohol use. For the model of “ever” cigarette smoking, 1,375 individuals were not included; for the model of frequency of cigarette smoking, 1,593 individuals were not included; for the model of duration of cigarette smoking, 1,553 individuals were not included; and for the model of pack-years, 1,681 individuals were not included.

b “Never smoker” was defined as never having smoked the product of interest. For example, a never smoker of cigarettes did not smoke cigarettes but may have smoked cigars or pipes. “Ever smoker” was defined as ever having smoked the product of interest, in a similar manner. For this reason, cigarette smoking odds ratios were adjusted for the duration of cigar smoking and the duration of pipe smoking.

cP for linear trend was obtained from modeling the continuous forms of the frequency, duration, and cumulative variables.

Table 3.

Odd Ratios for Cigar Smoking and Head and Neck Cancer in the International Head and Neck Cancer Epidemiology Consortium, 1981–2007

Variable Cigar Smoking
 
Never Cigarette Smoking
 
Ever Cigarette Smoking
 
No. of Cases No. of Controls ORa 95% CI No. of Cases No. of Controls ORb 95% CI 
Tobacco smokingc         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 Ever 171 155 2.54 1.93, 3.34 779 814 0.90 0.78, 1.02 
 Missing   15 14   
Smoking frequency (cigars per day)         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 1–10 130 136 1.99 1.47, 2.71 550 618 0.89 0.77, 1.03 
 >10 39 14 10.13 4.96, 20.67 184 155 0.99 0.74, 1.32 
 Missing   60 55   
  Ptrendd   <0.0001   0.62 
Smoking duration, years         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 1–10 19 52 0.91 0.51, 1.62 331 391 0.77 0.64, 0.92 
 11–20 18 19 2.42 1.18, 4.95 114 130 1.04 0.77, 1.42 
 21–30 18 24 1.92 0.95, 3.85 99 113 0.98 0.70, 1.36 
 31–40 43 29 3.78 2.18, 6.55 91 78 1.20 0.83, 1.72 
 >40 68 27 5.62 3.29, 9.60 123 87 1.04 0.73, 1.48 
 Missing   36 29   
  Ptrend    <0.0001   0.63 
Cumulative cigar-years         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 1–10 18 44 0.96 0.52, 1.77 162 245 0.71 0.56, 0.90 
 11–20 20 41 1.15 0.63, 2.11 226 228 0.99 0.79, 1.23 
 >20 127 64 4.87 3.36, 7.06 337 296 1.01 0.82, 1.25 
 Missing 10   69 59   
  Ptrend   <0.0001   0.05 
Variable Cigar Smoking
 
Never Cigarette Smoking
 
Ever Cigarette Smoking
 
No. of Cases No. of Controls ORa 95% CI No. of Cases No. of Controls ORb 95% CI 
Tobacco smokingc         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 Ever 171 155 2.54 1.93, 3.34 779 814 0.90 0.78, 1.02 
 Missing   15 14   
Smoking frequency (cigars per day)         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 1–10 130 136 1.99 1.47, 2.71 550 618 0.89 0.77, 1.03 
 >10 39 14 10.13 4.96, 20.67 184 155 0.99 0.74, 1.32 
 Missing   60 55   
  Ptrendd   <0.0001   0.62 
Smoking duration, years         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 1–10 19 52 0.91 0.51, 1.62 331 391 0.77 0.64, 0.92 
 11–20 18 19 2.42 1.18, 4.95 114 130 1.04 0.77, 1.42 
 21–30 18 24 1.92 0.95, 3.85 99 113 0.98 0.70, 1.36 
 31–40 43 29 3.78 2.18, 6.55 91 78 1.20 0.83, 1.72 
 >40 68 27 5.62 3.29, 9.60 123 87 1.04 0.73, 1.48 
 Missing   36 29   
  Ptrend    <0.0001   0.63 
Cumulative cigar-years         
 Never 1,567 6,824 1.00 Referent 10,525 9,759 1.00 Referent 
 1–10 18 44 0.96 0.52, 1.77 162 245 0.71 0.56, 0.90 
 11–20 20 41 1.15 0.63, 2.11 226 228 0.99 0.79, 1.23 
 >20 127 64 4.87 3.36, 7.06 337 296 1.01 0.82, 1.25 
 Missing 10   69 59   
  Ptrend   <0.0001   0.05 

Abbreviations: CI, confidence interval; OR, odds ratio.

a Cigar smoking odds ratios among never cigarette smokers were adjusted for sex, age, race, educational level, frequency of alcohol use (mL of ethanol per day), and duration of pipe smoking (in years). Individuals missing the main variable or 1 or more covariates were dropped from the models. The most frequently missing variable was frequency of alcohol use. For the model of “ever” smoking, 276 individuals were not included; for the model of frequency of smoking, 283 individuals were not included; for the model of duration of smoking, 285 individuals were not included; and for the model of cigar-years, 288 individuals were not included.

b Cigar smoking odds ratios among ever cigarette smokers were adjusted for sex, age, race, educational level, frequency of alcohol use (mL of ethanol per day), duration of cigarette smoking (in years), and duration of pipe smoking (in years). Individuals missing the main variable or 1 or more covariates were dropped from the models. The most frequently missing variable was frequency of alcohol use. For the model of “ever” smoking, 1,101 individuals were not included; for the model of frequency of smoking, 1,172 individuals were not included; for the model of duration of smoking, 1,127 individuals were not included; and for the model of cigar-years, 1,184 individuals were not included.

c “Never smoker” was defined as never having smoked the product of interest. For example, a never smoker of cigars did not smoke cigars but may have smoked pipes. “Ever smoker” was defined as ever having smoked the product of interest, in a similar manner. For this reason, cigar smoking odds ratios were adjusted for the duration of pipe smoking (and the duration of cigarette smoking among ever cigarette smokers).

dP for linear trend was obtained from modeling the continuous forms of the frequency, duration, and cumulative variables.

Table 4.

Odd Ratios for Pipe Smoking and Head and Neck Cancer in the International Head and Neck Cancer Epidemiology Consortium, 1981–2007

Variable Pipe Smoking
 
Never Cigarette Smoking
 
Ever Cigarette Smoking
 
No. of Cases No. of Controls ORa 95% CI No. of Cases No. of Controls ORb 95% CI 
Tobacco smokingc         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 Ever 138 123 2.08 1.55, 2.81 587 717 0.81 0.70, 0.93 
 Missing   17 16   
Smoking frequency (pipes per day)         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 1–10 118 107 2.11 1.53, 2.90 484 645 0.80 0.69, 0.93 
 >10 14 11 2.19 0.89, 5.40 69 37 1.00 0.60, 1.69 
 Missing 12   51 51   
  Ptrendd   0.0001   0.57 
Smoking duration, years         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 1–10 28 52 1.27 0.76, 2.14 345 432 0.79 0.66, 0.94 
 11–20 14 19 1.52 0.71, 3.23 64 120 0.77 0.54, 1.10 
 21–30 17 0.82 0.32, 2.13 59 83 0.70 0.47, 1.05 
 31–40 21 18 2.34 1.11, 4.91 41 39 1.01 0.61, 1.68 
 >40 66 17 5.81 3.15, 10.70 62 39 0.98 0.59, 1.63 
 Missing   33 20   
  Ptrend   <0.0001   0.16 
Cumulative pipe-years         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 1–10 20 39 1.31 0.72, 2.40 198 289 0.78 0.63, 0.97 
 11–20 19 36 0.89 0.47, 1.69 147 208 0.67 0.52, 0.86 
 >20 93 43 3.76 2.45, 5.76 205 185 1.04 0.81, 1.34 
 Missing 12   54 51   
  Ptrend   <0.0001   0.76 
Variable Pipe Smoking
 
Never Cigarette Smoking
 
Ever Cigarette Smoking
 
No. of Cases No. of Controls ORa 95% CI No. of Cases No. of Controls ORb 95% CI 
Tobacco smokingc         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 Ever 138 123 2.08 1.55, 2.81 587 717 0.81 0.70, 0.93 
 Missing   17 16   
Smoking frequency (pipes per day)         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 1–10 118 107 2.11 1.53, 2.90 484 645 0.80 0.69, 0.93 
 >10 14 11 2.19 0.89, 5.40 69 37 1.00 0.60, 1.69 
 Missing 12   51 51   
  Ptrendd   0.0001   0.57 
Smoking duration, years         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 1–10 28 52 1.27 0.76, 2.14 345 432 0.79 0.66, 0.94 
 11–20 14 19 1.52 0.71, 3.23 64 120 0.77 0.54, 1.10 
 21–30 17 0.82 0.32, 2.13 59 83 0.70 0.47, 1.05 
 31–40 21 18 2.34 1.11, 4.91 41 39 1.01 0.61, 1.68 
 >40 66 17 5.81 3.15, 10.70 62 39 0.98 0.59, 1.63 
 Missing   33 20   
  Ptrend   <0.0001   0.16 
Cumulative pipe-years         
 Never 1,612 7,022 1.00 Referent 10,888 10,044 1.00 Referent 
 1–10 20 39 1.31 0.72, 2.40 198 289 0.78 0.63, 0.97 
 11–20 19 36 0.89 0.47, 1.69 147 208 0.67 0.52, 0.86 
 >20 93 43 3.76 2.45, 5.76 205 185 1.04 0.81, 1.34 
 Missing 12   54 51   
  Ptrend   <0.0001   0.76 

Abbreviations: CI, confidence interval; OR, odds ratio.

a Pipe smoking odds ratios among never cigarette smokers were adjusted for sex, age, race, educational level, frequency of alcohol use (mL of ethanol per day), and duration of cigar smoking (in years). Individuals missing the main variable or 1 or more covariates were dropped from the models. The most frequently missing variable was frequency of alcohol use. For the model of “ever” smoking, 283 individuals were not included; for the model of smoking frequency, 294 individuals were not included; for the model of smoking duration, 285 individuals were not included; and for the model of pipe-years, 294 individuals were not included.

b Pipe smoking odds ratios among ever cigarette smokers were adjusted for sex, age, race, educational level, frequency of alcohol use (mL of ethanol per day), duration cigarette smoking (in years), and duration cigar smoking (in years). Individuals missing the main variable or 1 or more covariates were dropped from the models. The most frequently missing variable was frequency of alcohol use. For the model of “ever” smoking, 1,122 individuals were not included; for the model of smoking frequency, 1,170 individuals were not included; for the model of smoking duration, 1,131 individuals were not included; and for the model of pipe-years, 1,172 individuals were not included.

c “Never smoker” was defined as never having smoked the product of interest. For example, a never smoker of pipes did not smoke pipes but may have smoked cigars. “Ever smoker” was defined as ever having smoked the product of interest, in a similar manner. For this reason, pipe smoking odds ratios were adjusted for the duration of cigar smoking (and the duration of cigarette smoking among ever cigarette smokers).

dP for linear trend was obtained from modeling the continuous forms of the frequency, duration, and cumulative variables.

When we separately considered tumors in the oral cavity, pharynx, hypopharynx, oropharynx, larynx, and oral cavity/pharynx not otherwise specified, associations between cigarette smoking and laryngeal cancer and hypopharyngeal cancer resulted in odds ratios of the largest magnitudes (8.33 and 6.48, respectively) (Web Table 1 available at http://aje.oxfordjournals.org/). When subsite-specific estimates for ever cigar smoking and ever pipe smoking were stratified by ever cigarette smoking, the odds ratios were consistently elevated among never cigarette smokers but not among ever smokers across subsites (though several estimates were imprecise because of low cell counts and should therefore be interpreted with caution) (Web Table 1). Among never cigarette smokers, associations between cigar and pipe smoking and laryngeal cancer again resulted in odds ratios of the largest magnitudes (6.31 and 3.53, respectively). For oropharyngeal cancers, the odds ratios were 2.31 (95% CI: 1.54, 3.45) for ever cigar smoking and 1.65 (95% CI: 1.04, 2.60) for ever pipe smoking among never cigarette smokers. When we restricted analysis to tumors with squamous cell carcinoma histology, the odds ratios were similar to the overall head and neck cancer estimates for each tobacco product (Web Table 1). The odds ratio for cigarette smoking and head and neck squamous cell carcinoma was 3.37 (95% CI: 3.12, 3.64). Likewise, stratification of squamous cell carcinoma–specific estimates for cigar and pipe smoking by cigarette smoking again resulted in elevated odds ratios among never cigarette smokers (OR = 2.77, 95% CI: 2.08, 3.71 and OR = 1.89, 95% CI: 1.36, 2.63, respectively).

The odds ratios for cigarette smoking were elevated in each region considered, but appeared to be lower for North American smokers than for smokers in Europe and South America (Web Table 1). Cigarette smokers aged 45 years and older appeared to have a higher risk than cigarette smokers under 45 years of age (Web Table 1). Effect estimates for cigar and pipe smoking among never cigarette smokers appeared to follow similar patterns by region and age; however, some estimates were imprecise because of low cell counts (Web Table 1). Among women, odds ratios were elevated, but estimates were based on relatively few female smokers, especially ever cigar and ever pipe smokers who never smoked cigarettes, as evidenced by confidence intervals that were much wider than those among males, and should therefore be interpreted with caution (Web Table 1).

Analyses of exclusive cigarette, cigar, and pipe smokers yielded elevated odds ratios with similar magnitudes. Compared with never smoking cigarettes, cigars, or pipes, the adjusted odds ratios were 3.93 (95% CI: 3.67, 4.22) for smoking only cigarettes, 3.49 (95% CI: 2.58, 4.73) for smoking only cigars, and 3.71 (95% CI: 2.59, 5.33) for smoking only pipes (Table 5). Likewise, odds ratios were elevated for each type of smoking tobacco when dominant and predominant smoking habits were considered (Table 5). Joint effects from smoking more than 1 type of tobacco were also elevated compared with never smokers (Table 5) but not when compared with only cigarette smokers (data not shown). Because of small counts across strata in many of these analyses, we were unable to consider the duration and frequency of use among exclusive users.

Table 5.

Odd Ratios for Exclusive/Joint and Dominant/Predominant Cigarette, Cigar, and Pipe Smoking and Head and Neck Cancer in the International Head and Neck Cancer Epidemiology Consortium, 1981–2007

Variablea No. of Cases No. of Controls ORb 95% CI 
Exclusive/joint tobacco smoking     
 Never smoker 1,567 7,154 1.00 Referent 
 Cigarette, cigar, and pipe 211 277 2.42 1.96, 2.99 
 Cigarette and cigar 566 541 3.05 2.61, 3.56 
 Cigarette and pipe 371 445 2.94 2.49, 3.48 
 Cigar and pipe 61 52 3.44 2.29, 5.17 
 Cigarette only 10,880 9,996 3.93 3.67, 4.22 
 Cigar only 110 103 3.49 2.58, 4.73 
 Pipe only 77 71 3.71 2.59, 5.33 
 Missing 92 52   
Dominant/predominant tobacco smoking     
 Never smoker 1,567 7,154 1.00 Referent 
 Predominantly cigarette 659 736 2.88 2.51, 3.30 
 Only cigarette 10,755 9,895 3.94 3.67, 4.23 
 Predominantly cigar 131 111 3.22 2.38, 4.35 
 Only cigar 107 97 3.56 2.62, 4.86 
 Predominantly pipe 48 50 2.93 1.89, 4.55 
 Only pipe 75 70 3.67 2.55, 5.28 
 Less than 66.6% for any smoking product 237 261 2.72 2.20, 3.35 
 Missing 356 317   
Variablea No. of Cases No. of Controls ORb 95% CI 
Exclusive/joint tobacco smoking     
 Never smoker 1,567 7,154 1.00 Referent 
 Cigarette, cigar, and pipe 211 277 2.42 1.96, 2.99 
 Cigarette and cigar 566 541 3.05 2.61, 3.56 
 Cigarette and pipe 371 445 2.94 2.49, 3.48 
 Cigar and pipe 61 52 3.44 2.29, 5.17 
 Cigarette only 10,880 9,996 3.93 3.67, 4.22 
 Cigar only 110 103 3.49 2.58, 4.73 
 Pipe only 77 71 3.71 2.59, 5.33 
 Missing 92 52   
Dominant/predominant tobacco smoking     
 Never smoker 1,567 7,154 1.00 Referent 
 Predominantly cigarette 659 736 2.88 2.51, 3.30 
 Only cigarette 10,755 9,895 3.94 3.67, 4.23 
 Predominantly cigar 131 111 3.22 2.38, 4.35 
 Only cigar 107 97 3.56 2.62, 4.86 
 Predominantly pipe 48 50 2.93 1.89, 4.55 
 Only pipe 75 70 3.67 2.55, 5.28 
 Less than 66.6% for any smoking product 237 261 2.72 2.20, 3.35 
 Missing 356 317   

Abbreviations: CI, confidence interval; OR, odds ratio.

a For the exclusive/joint model, “never smoker” was defined as never having smoked cigarettes, cigars, and pipes; joint and exclusive smokers of tobacco products were defined as ever having smoked the specified tobacco product(s) and never having smoked any other tobacco products. For the dominant/predominant model, “never smoker” was defined as having 0 g of lifetime consumption of cigarettes, cigars, and pipes; predominant smokers of a given product were defined as the lifetime consumption of the specified smoking product (in grams) divided by the sum of lifetime consumption of all smoking products (in grams) of between 66.6% and 99.9%; and dominant (only) smokers of a given product were defined as lifetime consumption of specified smoking product (in grams) divided by the sum of lifetime consumption of all smoking products (in grams) of 100%.

b Odds ratios were adjusted for sex, age, race, educational level, and frequency of alcohol use (mL of ethanol per day). A total of 1,319 individuals were not included in the exclusive/joint tobacco smoking model for missing the main variable or 1 or more covariates. A total of 1,788 individuals were not included in the dominant/predominant model for missing the main variable or 1 or more covariates. The most frequently missing covariate was frequency of alcohol use. (Additional footnotes are available in Web material at http://aje.oxfordjournals.org/.)

DISCUSSION

We found that cigarette smokers had elevated odds ratios for head and neck cancers, consistent with extensive literature (3). Given the strong association between cigarette smoking and head and neck cancers and the large proportion of cases and controls who had smoked cigarettes, simply adjusting the cigar and pipe smoking models for the duration of cigarette smoking would not have adequately controlled for the confounding influence of cigarette smoking. Therefore, we stratified estimates by ever cigarette smoking to obtain cigar and pipe smoking odds ratios among individuals not influenced by active cigarette smoking. Among never cigarette smokers, we observed an increased risk associated with ever cigar and ever pipe smoking. The risk of head and neck cancers also increased with increasing levels of frequency and duration of cigar or pipe smoking among never cigarette smokers. Moreover, elevated odds ratios associated with exclusive smoking of cigars or pipes compared with those of a reference group of never smokers of cigarettes, cigars, or pipes suggest that both products independently contribute to head and neck cancer risk. Similarly, analyses considering predominant cigar smokers and predominant pipe smokers also resulted in increased odds ratios compared with those of a reference group of never smokers of any product.

Although we observed strong associations between cigar and pipe smoking and head and neck cancer risk in the absence of cigarette smoking, the added impact of cigar and pipe smoking in the presence of cigarette smoking was not discernible. Specifically, we noted near null odds ratios for cigar and pipe smoking among ever cigarette smokers in stratified analyses. Further, the odds ratios for the joint smoking of products were not appreciably different from those for the exclusive smoking of single products. One explanation for null cigar and pipe smoking results among ever cigarette smokers may be residual confounding attributed to cigarette smoking. Adjustment by other smoking metrics did not materially change the interpretation. Misclassification bias or unmeasured confounders may also explain the null results for cigar and pipe smoking among ever cigarette smokers. Another explanation may be that smokers of more than 2 products differ from smokers of only 1 product with regard to duration and frequency of use of each product. Among individuals who smoke cigarettes and cigars or pipes, the greater frequency of cigarette smoking may make cigarette smoke the predominant source of carcinogen exposure.

The finding of elevated head and neck cancer risk among pipe and cigar smokers is consistent with those of previous studies, including both individual studies in this pooled analysis (14–18) and non–INHANCE Consortium studies (19–28), as well as studies on mortality (mostly disease-specific mortality) (29–34). Among previous studies that were not included in the INHANCE Consortium, effect estimates for the association between cigar and/or pipe smoking and head and neck cancers ranged from approximately null to more than 5.0 (19–28). Among studies of exclusive cigar and/or pipe smoking, 2 reported elevated odds ratios of magnitudes similar to our estimates (i.e., ORs of approximately 2.5–3.5) (20, 22). Most recently, a study that used data from the European Prospective Investigation into Cancer Nutrition cohort noted elevated risk of upper aerodigestive tract cancers (head and neck cancers and esophageal cancer) associated with exclusive cigar smoking (hazard ratio = 4.0, 95% CI: 1.7, 9.4) and exclusive pipe smoking (hazard ratio = 1.7, 95% CI: 0.6, 4.4), although estimates were based on relatively few cases (7 cases exclusively smoked cigars and 5 exclusively smoked pipes) (28).

As detailed in an International Agency for Research on Cancer monograph, tobacco smoke contains numerous carcinogens (3). Although cigar and pipe smoking are not as common as cigarette smoking, cigar and pipe products contain higher average weights of tobacco than do cigarette products (3). Machine-smoking studies have further suggested that delivered doses of tar, nicotine, carbon monoxide, benzene, benzo[a]pyrene, and tobacco-specific nitrosamines from cigar and pipe smoking are comparable to, and in some instances higher than, those from cigarette smoking (3). Given similarities between delivered carcinogens across tobacco smoking products and the direct exposure of the oral cavity, pharynx, and larynx to tobacco smoke, it seems reasonable that associations between cigar or pipe smoking and head and neck cancer risk would be similar in magnitude to those between cigarette smoking and head and neck cancer risk.

The major strength of our study was the evaluation of multiple tobacco products by using a large international data set. The analysis of exclusive and predominant smokers of cigars and pipes was another key contribution. Although we were unable to analyze the frequency and duration of smoking among exclusive pipe and cigar smokers because of small cell counts, it should be noted that this analysis had the largest sample size to date. Most previous studies that evaluated cigar and pipe smoking and head and neck cancer risk (and were not included in the INHANCE Consortium) had sample sizes of approximately 100 to a few thousand cases and controls, with approximately 10–100 cases and controls who ever or exclusively smoked cigars or pipes (19–28).

Another asset of our study was detailed information on important covariates. In addition to stratifying cigar and pipe analyses by cigarette smoking, we adjusted all estimates for the frequency of alcohol drinking. Previous INHANCE Consortium studies reported strong dose-response relationships between the frequency of alcohol drinking and head and neck cancers, especially among never cigarette smokers (7). Further, all models were adjusted and/or stratified by demographic factors including age, gender, race, and educational level. Subsite-specific models were also considered, including estimates specific to oropharyngeal cancers. Information on human papillomavirus status is currently not available in INHANCE Consortium data. Future studies that incorporate longitudinal designs with serial assessment of confounders would further strengthen conclusions regarding cigar and pipe smoking and head and neck cancers.

Because information on grams of smoking tobacco consumed was not available, we had to use conversions based on estimates of the average weight of tobacco in each smoking product for lifetime consumption analyses. It is also important to note that cumulative measures of smoking (i.e., pack-years, cigar-years, and pipe-years) are based on the total duration and average frequency of smoking of the given product and may not reflect fluctuations in the duration and frequency of smoking across the life span. Furthermore, cumulative measures do not account for the amount of delivered carcinogen in each tobacco product nor for unique habits in how products are smoked. Finally, main analyses included head and neck cancers with histologies other than squamous cell carcinomas (approximately 87.6% of cases with reported histology were squamous cell carcinomas, and the remaining 12.4% encompassed other histologies). However, odds ratios for ever smoking each product and head and neck cancers restricted to squamous cell carcinomas were similar to the estimates for the overall analyses.

Our findings of an increased risk of head and neck cancers among cigarette smokers are consistent with those of the previous literature. We also found increased risk of head and neck cancers among cigar smokers and pipe smokers, with the magnitude of the associations for cigar or pipe smoking similar to that for cigarette smoking. The results of the present study provide more precise estimates of the risks of head and neck cancers associated with pipes and cigars, especially among never cigarette smokers, considerably strengthening the evidence that pipe smoking and cigar smoking are risk factors for head and neck cancers.

ACKNOWLEDGMENTS

Author affiliations: Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, North Carolina (Annah Wyss, Andrew F. Olshan); Department of Family and Preventive Medicine, School of Medicine, University of Utah, Salt Lake City, Utah (Mia Hashibe, Yuan-Chin Amy Lee); Institute of Population Health Sciences, National Health Research Institutes, Miaoli, Taiwan (Shu-Chun Chuang); Department of Epidemiology, School of Public Health, University of California Los Angeles, Los Angeles, California (Zuo-Feng Zhang); Department of Otolaryngology, New York Medical College, New York, New York (Guo-Pei Yu); Medical Informatics Center, Peking University, Beijing, China (Guo-Pei Yu); Division of Cancer Control and Population Sciences, National Cancer Institute, National Institutes of Health, Bethesda, Maryland (Deborah M. Winn); Department of Epidemiology, Division of Cancer Prevention and Population Sciences, University of Texas MD Anderson Cancer Center, Houston, Texas (Qingyi Wei, Erich M. Sturgis); Epidemiology and Biostatistics Unit, National Cancer Institute of Aviano, Aviano, Italy (Renato Talamini, Luigino Dal Maso); Department of Epidemiology, Institute of Occupational Medicine, Lodz, Poland (Neonila Szeszenia-Dabrowska); Department of Head and Neck Surgery, Division of Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas (Erich M. Sturgis); Department of Epidemiology, College of Public Health, University of Iowa, Iowa City, Iowa (Elaine Smith); Department of Epidemiology and Prevention, Institute of Carcinogenesis, N. N. Blokhin Cancer Research Center, Moscow, Russia (Oxana Shangina); Department of Epidemiology, School of Public Health, University of Washington, and Program in Epidemiology, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington (Stephen M. Schwartz, Chu Chen); Biostatistics and Epidemiology Service, Department of Otolaryngology, The New York Eye and Ear Infirmary, New York, New York (Stimson Schantz); Department of Environmental Epidemiology, National Institute of Environmental Health, Budapest, Hungary (Peter Rudnai); Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, Maryland (Mark P. Purdue); Department of Preventive Medicine, Faculty of Medicine, University of São Paulo, São Paulo, Brazil (Jose Eluf-Neto); Department of Public Health Sciences, Penn State College of Medicine, Hershey, Pennsylvania (Joshua Muscat); Departments of Epidemiology and Environmental Health Sciences, School of Public Health and Comprehensive Cancer Center, University of Michigan, Ann Arbor, Michigan (Hal Morgenstern); Head and Neck Service, Faculty of Medicine, University of São Paulo, São Paulo, Brazil (Pedro Michaluart, Jr.); Federal University of Pelotas, Pelotas, Brazil (Ana Menezes); Angel H. Roffo Institute of Oncology, University of Buenos Aires, Buenos Aires, Argentina (Elena Matos); Carol Davila University of Medicine and Pharmacy, Bucharest, Romania (Ioan Nicolae Mates); Department of Cancer Epidemiology and Prevention, The Maria Skłodowska-Curie Memorial Cancer Centre and Institute of Oncology, Warsaw, Poland (Jolanta Lissowska); Cancer Epidemiology Unit and Vaud Cancer Registry, University Institute of Social and Preventive Medicine, Vaudois University Hospital Center and University of Lausanne, Lausanne, Switzerland (Fabio Levi); Department of Pharmaceutical Sciences, College of Pharmacy, Washington State University, Spokane, Washington (Philip Lazarus); Department of Epidemiology, The Mario Negri Institute for Pharmacological Research, and Department of Clinical Sciences and Community Health, University of Milan, Milan, Italy (Carlo La Vecchia); Department of Epidemiology, National School of Public Health, Oswaldo Cruz Foundation, Rio de Janeiro, Brazil (Sergio Koifman); Prevention and Implementation Group, International Agency for Research on Cancer, Lyon, France (Rolando Herrero); Division of Epidemiology, Department of Population Health, New York University School of Medicine, New York, New York (Richard B. Hayes); Infections and Cancer Epidemiology Group, International Agency for Research on Cancer, Lyon, France (Silvia Franceschi); Department of Epidemiology, Faculty of Public Health, University of São Paulo, São Paulo, Brazil (Victor Wünsch-Filho); Institute of Oncology and Radiobiology, Havana, Cuba (Leticia Fernandez); Regional Authority of Public Health Banská Bystrica, Banská Bystrica, Slovakia (Eleonora Fabianova); Clinical Hospital of Porto Alegre, Porto Alegre, Brazil (Alexander W. Daudt); International Prevention Research Institute, Lyon, France (Maria Paula Curado, Paolo Boffetta); Catalan Institute of Oncology, L'Hospitalet de Llobregat, Spain (Xavier Castellsague); Molecular Biology Laboratory, Heliópolis Hospital, São Paulo, Brazil (Marcos Brasilino de Carvalho); Department of Head and Neck Surgery–Otorhinolaryngology, A. Gemelli Hospital, Catholic University of the Sacred Heart, Rome, Italy (Gabriella Cadoni); Institute of Hygiene, Catholic University of the Sacred Heart and Institute for Research, Hospitalization, and Health Care, San Raffaele Pisana, Rome, Italy (Stefania Boccia); Genetic Epidemiology Group, International Agency for Research on Cancer, Lyon, France (Paul Brennan); The Tisch Cancer Institute, Mount Sinai School of Medicine, New York, New York (Paolo Boffetta).

This work was supported by a Union for International Cancer Control International Cancer Technology Transfer Fellowship, by the National Cancer Institute (grants T32-CA09330 and R03 CA113157), and by the National Institute of Environmental Health Sciences (grant T32ES007018).

We thank Geraldine Dominiak for her assistance with the data analysis and Dr. Patrick Bradshaw, Nikhil Khankari, and Dr. David Richardson for their assistance with hierarchical models.

Conflict of interest: none declared.

REFERENCES

1
Parkin
DM
Bray
F
Ferlay
J
, et al.  . 
Global cancer statistics, 2002
CA Cancer J Clin
 , 
2005
, vol. 
55
 
2
(pg. 
74
-
108
)
2
Curado
MP
Hashibe
M
Recent changes in the epidemiology of head and neck cancer
Curr Opin Oncol
 , 
2009
, vol. 
21
 
3
(pg. 
194
-
200
)
3
International Agency for Research on Cancer
IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Volume 83: Tobacco Smoke and Involuntary Smoking
 , 
2004
Lyon, France
International Agency for Research on Cancer
4
Conway
DI
Hashibe
M
Boffetta
P
, et al.  . 
Enhancing epidemiologic research on head and neck cancer: INHANCE—the International Head and Neck Cancer Epidemiology Consortium
Oral Oncol
 , 
2009
, vol. 
45
 
9
(pg. 
743
-
746
)
5
Benhamou
S
Tuimala
J
Bouchardy
C
, et al.  . 
DNA repair gene XRCC2 and XRCC3 polymorphisms and susceptibility to cancers of the upper aerodigestive tract
Int J Cancer
 , 
2004
, vol. 
112
 
5
(pg. 
901
-
904
)
6
Balaram
P
Sridhar
H
Rajkumar
T
, et al.  . 
Oral cancer in southern India: the influence of smoking, drinking, paan-chewing and oral hygiene
Int J Cancer
 , 
2002
, vol. 
98
 
3
(pg. 
440
-
445
)
7
Hashibe
M
Brennan
P
Benhamou
S
, et al.  . 
Alcohol drinking in never users of tobacco, cigarette smoking in never drinkers, and the risk of head and neck cancer: pooled analysis in the International Head and Neck Cancer Epidemiology Consortium
J Natl Cancer Inst
 , 
2007
, vol. 
99
 
10
(pg. 
777
-
789
)
8
World Health Organization
International Statistical Classification of Diseases, Injuries, and Causes of Death
 , 
1977
Geneva, Switzerland
World Health Organization
 
Ninth Revision
9
World Health Organization
International Statistical Classification of Diseases, Injuries, and Causes of Death, Tenth Revision
 , 
1992–1994
Geneva, Switzerland
World Health Organization
10
Percy
C
Van Holten
V
Muir
C
International Classification of Diseases for Oncology
 , 
1990
2nd ed
Geneva, Switzerland
World Health Organization
11
SAS/STAT User's Guide, Version 8: The NLMixed Procedure
 , 
1999
Cary, NC
SAS Institute, Inc
12
Berthiller
J
Lee
YC
Boffetta
P
, et al.  . 
Marijuana smoking and the risk of head and neck cancer: pooled analysis in the INHANCE Consortium
Cancer Epidemiol Biomarkers Prev
 , 
2009
, vol. 
18
 
5
(pg. 
1544
-
1551
)
13
Mayne
ST
Morse
DE
Winn
DM
Schottenfeld
D
Fraumeni
JF
Cancers of the oral cavity and pharynx
Cancer Epidemiology and Prevention
 , 
2006
3rd ed
NewYork, NY
Oxford University Press
(pg. 
674
-
696
)
14
La Vecchia
C
Bosetti
C
Negri
E
, et al.  . 
Cigar smoking and cancers of the upper digestive tract [correspondence]
J Natl Cancer Inst
 , 
1998
, vol. 
90
 
21
pg. 
1670
 
15
Randi
G
Scotti
L
Bosetti
C
, et al.  . 
Pipe smoking and cancers of the upper digestive tract
Int J Cancer
 , 
2007
, vol. 
121
 
9
(pg. 
2049
-
2051
)
16
Franceschi
S
Talamini
R
Barra
S
, et al.  . 
Smoking and drinking in relation to cancers of the oral cavity, pharynx, larynx, and esophagus in northern Italy
Cancer Res
 , 
1990
, vol. 
50
 
20
(pg. 
6502
-
6507
)
17
Franceschi
S
Barra
S
La Vecchia
C
, et al.  . 
Risk factors for cancer of the tongue and the mouth. A case-control study from northern Italy
Cancer
 , 
1992
, vol. 
70
 
9
(pg. 
2227
-
2233
)
18
Blot
WJ
McLaughlin
JK
Winn
DM
, et al.  . 
Smoking and drinking in relation to oral and pharyngeal cancer
Cancer Res
 , 
1988
, vol. 
48
 
11
(pg. 
3282
-
3287
)
19
Garrote
LF
Herrero
R
Reyes
RM
, et al.  . 
Risk factors for cancer of the oral cavity and oro-pharynx in Cuba
Br J Cancer
 , 
2001
, vol. 
85
 
1
(pg. 
46
-
54
)
20
Kabat
GC
Chang
CJ
Wynder
EL
The role of tobacco, alcohol use, and body mass index in oral and pharyngeal cancer
Int J Epidemiol
 , 
1994
, vol. 
23
 
6
(pg. 
1137
-
1144
)
21
Iribarren
C
Tekawa
IS
Sidney
S
, et al.  . 
Effect of cigar smoking on the risk of cardiovascular disease, chronic obstructive pulmonary disease, and cancer in men
N Engl J Med
 , 
1999
, vol. 
340
 
23
(pg. 
1773
-
1780
)
22
Mashberg
A
Boffetta
P
Winkelman
R
, et al.  . 
Tobacco smoking, alcohol drinking, and cancer of the oral cavity and oropharynx among US veterans
Cancer
 , 
1993
, vol. 
72
 
4
(pg. 
1369
-
1375
)
23
Zheng
TZ
Boyle
P
Hu
HF
, et al.  . 
Tobacco smoking, alcohol consumption, and risk of oral cancer: a case-control study in Beijing, People's Republic of China
Cancer Causes Control
 , 
1990
, vol. 
1
 
2
(pg. 
173
-
179
)
24
Spitz
MR
Fueger
JJ
Goepfert
H
, et al.  . 
Squamous cell carcinoma of the upper aerodigestive tract. A case comparison analysis
Cancer
 , 
1988
, vol. 
61
 
1
(pg. 
203
-
208
)
25
Stockwell
HG
Lyman
GH
Impact of smoking and smokeless tobacco on the risk of cancer of the head and neck
Head Neck Surg
 , 
1986
, vol. 
9
 
2
(pg. 
104
-
110
)
26
Williams
RR
Horm
JW
Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: interview study from the Third National Cancer Survey
J Natl Cancer Inst
 , 
1977
, vol. 
58
 
3
(pg. 
525
-
547
)
27
Spitz
MR
Fueger
JJ
Halabi
S
, et al.  . 
Mutagen sensitivity in upper aerodigestive tract cancer: a case-control analysis
Cancer Epidemiol Biomarkers Prev
 , 
1993
, vol. 
2
 
4
(pg. 
329
-
333
)
28
McCormack
VA
Agudo
A
Dahm
CC
, et al.  . 
Cigar and pipe smoking and cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC)
Int J Cancer
 , 
2010
, vol. 
127
 
10
(pg. 
2402
-
2411
)
29
Henley
SJ
Thun
MJ
Chao
A
, et al.  . 
Association between exclusive pipe smoking and mortality from cancer and other diseases
J Natl Cancer Inst
 , 
2004
, vol. 
96
 
11
(pg. 
853
-
861
)
30
Shapiro
JA
Jacobs
EJ
Thun
MJ
Cigar smoking in men and risk of death from tobacco-related cancers
J Natl Cancer Inst
 , 
2000
, vol. 
92
 
4
(pg. 
333
-
337
)
31
Carstensen
JM
Pershagen
G
Eklund
G
Mortality in relation to cigarette and pipe smoking: 16 years’ observation of 25,000 Swedish men
J Epidemiol Community Health
 , 
1987
, vol. 
41
 
2
(pg. 
166
-
172
)
32
Kahn
HA
National Cancer Institute Monograph: Volume 19: The Dorn Study of Smoking and Mortality Among US Veterans: Report on Eight and One-Half Years of Observation
 , 
1966
Bethesda, MD
National Cancer Institute
33
Hammond
EC
Horn
D
Smoking and death rates: report on forty-four months of follow-up of 187,783 men. 2. Death rates by cause
J Am Med Assoc
 , 
1958
, vol. 
166
 
11
(pg. 
1294
-
1308
)
34
Zahm
SH
Heineman
EF
Vaught
JB
Soft tissue sarcoma and tobacco use: data from a prospective cohort study of United States veterans
Cancer Causes Control
 , 
1992
, vol. 
3
 
4
(pg. 
371
-
376
)

Author notes

Abbreviations: CI, confidence interval; INHANCE, International Head and Neck Cancer Epidemiology; OR, odds ratio.