Pebbles Fagan, Tobacco Control Research Branch, Behavioral Research Program, Division of Cancer Control and Population Sciences, National Cancer Institute, Executive Plaza North, Room 4042, 6130 Executive Blvd, MSC 7337, Bethesda, MD 20892-7337, USA. E-mail: email@example.com
Aims Few frameworks have addressed work-force diversity, inequities and inequalities as part of a comprehensive approach to eliminating tobacco-related health disparities. This paper summarizes the literature and describes the known disparities that exist along the tobacco disease continuum for minority racial and ethnic groups, those living in poverty, those with low education and blue-collar and service workers. The paper also discusses how work-force diversity, inequities in research practice and knowledge allocation and inequalities in access to and quality of health care are fundamental to addressing disparities in health.
Methods We examined the available scientific literature and existing public health reports to identify disparities across the tobacco disease continuum by minority racial/ethnic group, poverty status, education level and occupation.
Findings Results indicate that differences in risk indicators along the tobacco disease continuum do not explain fully tobacco-related cancer consequences among some minority racial/ethnic groups, particularly among the aggregate groups, blacks/African Americans and American Indians/Alaska Natives. The lack of within-race/ethnic group data and its interactions with socio-economic factors across the life-span contribute to the inconsistency we observe in the disease causal paradigm.
Conclusions More comprehensive models are needed to understand the relationships among disparities, social context, diversity, inequalities and inequities. A systematic approach will also help researchers, practitioners, advocates and policy makers determine critical points for interventions, the types of studies and programs needed and integrative approaches needed to eliminate tobacco-related disparities.
Tobacco is the leading cause of preventable death in the United States and contributes to increased incidence, morbidity and mortality from cancer, heart disease, stroke, complications of pregnancy and respiratory illness [1–3]. Approximately 438 000 people die annually from tobacco-related diseases , and some US minority racial and ethnic populations bear a disproportionate burden of tobacco-related health disparities . Tobacco-related health disparities (TRHD) are differences in exposure to tobacco, tobacco use initiation, current use, the number of cigarettes smoked per day (cpd), quitting, treatment, relapse and the subsequent health consequences among specific population groups, and include differences in capacity and infrastructure as well as access to resources . Some populations may not experience disparities along the entire ‘tobacco disease continuum’. However, others may experience severe and devastating effects along the entire causal pathway to diseases such as cancer [1,6,7].
The independent investigation of differences in the prevalence of tobacco use and exposure and the incidence and mortality of diseases has produced limited progress in eliminating tobacco-related diseases and deaths. For example, previous studies suggest that higher smoking prevalence rates explain higher rates of tobacco-related cancers among black/African American men compared to white men . Historically, the differences in smoking prevalence between black/African American men and white men were small, such that for the past 10 years rates have been fairly similar. Black/African American adolescents have historically had substantially lower smoking rates than white youth [8,9] and experience delayed onset of smoking [10,11]. Black/African American adults and adolescents smoke substantially fewer cigarettes per day, and adults have lower quit rates  and younger age of diagnosis of lung cancer compared to whites . Thus, existing tobacco data do not explain why black/African American men have the highest incidence and death rates from lung/bronchus, oral/pharynx, pancreatic, esophageal and larynx cancers . Black/African American women have historically had similar smoking rates to white women and rates lower than American Indian/Alaska Native women. However, lung cancer incidence is increasing among black/African American women and they have the highest incidence and second highest lung/bronchus death rates of all women (see Tables 1 and 2). New paradigms for examining disparities may help to increase our knowledge on how to eliminate disparities in minority racial/ethnic groups.
Table 1. SEER tobacco-related cancer death rates and trends by sex and race/ethnicity, 1992–2002.
Table adapted from table 4 printed in Edwards et al.  with permissions from the author and Oxford University Press. Sources of data are the Surveillance, Epidemiology, and End Results (SEER) Program registries that include Connecticut, Hawaii, Iowa, Utah, and New Mexico; the metropolitan areas of San Francisco, Detroit, Atlanta, Seattle-Puget Sound, San Jose-Monterey and Los Angeles; rural Georgia and Alaska Natives in Alaska (i.e. SEER 13). AC = absolute change; APC = annual percentage change; N/A = not available. *Data for Hispanics/Latino/Hispanics excludes cases diagnosed in Detroit, Hawaii, Alaska Natives and rural Georgia.†Rates are per 100 000 people and are age-adjusted to the 2000 US Standard Population—using 19 age groups, with data provided from US Bureau of the Census, Current Population Reports, Series P25–1130.‡APC is based on rates that were age-adjusted to the 2000 US. Standard Population—using 19 age groups, with data provided from US Bureau of the Census, Current Population Reports, Series P25–1130) (39).§AC was calculated as the difference in the age-adjusted rate for 2002 minus age-adjusted rate for 1992.¶Statistic could not be calculated. APC based on fewer than 10 cases for at least 1 year within the time interval.**APC is statistically significantly different from zero (two-sided P < 0.05).
Table adapted from Table 3 printed in Edwards et al.  with permissions from the author and Oxford University Press. Sources of data are the Surveillance, Epidemiology, and End Results (SEER) Program registries that include Connecticut, Hawaii, Iowa, Utah and New Mexico; the metropolitan areas of San Francisco, Detroit, Atlanta, Seattle-Puget Sound, San Jose-Monterey and Los Angeles; rural Georgia and Alaska Natives in Alaska (i.e. SEER 13). AC = absolute change; APC = annual percentage change; N/A = not available. *Data for Hispanics/Latino/Hispanics excludes cases diagnosed in Detroit, Hawaii, Alaska Natives and rural Georgia.†Rates are per 100 000 people and are age-adjusted to the 2000 US Standard Population—using 19 age groups, with data provided from US Bureau of the Census, Current Population Reports, Series P25–1130.‡APC is based on rates that were age-adjusted to the 2000 US Standard Population—using 19 age groups, with data provided from US Bureau of the Census, Current Population Reports, Series P25–1130) (39).§AC was calculated as the difference in the age-adjusted rate for 2002 minus age-adjusted rate for 1992.¶APC is statistically significantly different from zero (two-sided P < 0.05).**Statistic could not be calculated. APC based on fewer than 10 cases for at least 1 year during the time interval.
Few frameworks have addressed work-force diversity, inequities and inequalities as part of a comprehensive approach to eliminating tobacco-related health disparities (see Fig. 1). In fact, diversity and disparities have often been approached separately, although a relationship exists. Diversity, which includes integration and inclusion, can create helpful environments that increase researchers' understanding of disparities. National reports have suggested that ‘increasing racial and ethnic diversity among health professionals is important because the evidence indicates that diversity is associated with improved access to care for racial and ethnic minority patients . . .’. Addressing inequities, injustice or unfairness in science may help to improve scientific practice and the capacity to address understudied research questions. Inequalities, the quality of being unequal, exist in the quality of and access to care, patient experiences and opportunities for preventive care . Addressing inequalities may increase minority and racial and ethnic groups' access to cessation and cancer treatment services. Approaches that address simultaneously TRHD, work-force diversity and empowerment, inequities in research and inequalities in access to care may provide the paradigm shift needed to accelerate progress in reducing disparities among minority racial and ethnic groups (see Fig. 1).
This paper summarizes the known disparities that exist along the tobacco disease continuum for minority racial and ethnic groups, those living in poverty, those with low education and blue-collar and service workers. This paper also describes how addressing diversity, inequities in research and inequalities in care in tandem with disparities may help us reduce and eliminate TRHD. Minority racial and ethnic groups have suffered traditionally from or experienced social, cultural, educational, political and economic circumstances that greatly impact disparities . Minority racial and ethnic groups have a long history of discrimination, oppression and social marginalization within the United States, and the fact that health status varies by race/ethnicity is partly a consequence of history and social and cultural systems . Race and ethnicity may interact with socio-economic indicators such as poverty, education and occupation to increase or decrease disparities  at any point along the tobacco disease continuum. Minority racial and ethnic groups are also under-represented in the health research work-force, and there are inequities in research that impact our understanding of the multiple disparities from which these groups suffer. Understanding the relationship between TRHD among minority racial/ethnic groups and social and structural systems that impact disparities will help researchers, practitioners, advocates and policy makers determine optimal points along the tobacco disease continuum to intervene and integrative approaches needed to address this systemic epidemic.
We searched the scientific literature using PubMed and examined major scientific reports (e.g. Surgeon General Reports, National Cancer Institute monographs and reports, Agency for Health Care Quality and Research, Morbidity and Mortality Weekly Reports, US Census Bureau) to delineate known disparities along the tobacco disease continuum, including data on second-hand smoke or environmental tobacco smoke (ETS), tobacco use initiation, current use, number of cigarettes smoked per day, quitting, treatment, relapse and tobacco-related cancer morbidity and mortality. We reviewed existing national survey data and state and local survey data when national data were not available for small racial and ethnic groups. National data used to report estimates of tobacco use [i.e. National Health Interview Survey (NIHS), the Youth Risk Behavioral Survey (YRBS), Monitoring the Future Survey (MTF), the National Survey of Drug Use and Health (NSDUH)] were reviewed for the most recent data and trends. Data from the Centers for Disease Control and Prevention National Program of Cancer Registry and the National Cancer Institute's Surveillance Epidemiology and End Results Program were used to report tobacco-related cancers. Because eliminating tobacco use would almost eliminate lung cancer, the leading cause of cancer deaths (see Table 1), and tobacco is related causally to 10 other cancers, we used tobacco-related cancers as our main indicators of disease consequences. In the following sections, we summarize the known disparities for each minority racial/ethnic group and by socio-economic indicators, and discuss the roles of diversity, inequity and inequalities in reducing disparities.
A ‘disparity’ is defined as the condition or fact of being unequal in age, rank or degree , and is used to describe unequal health conditions or indicators (e.g. tobacco use initiation, ETS, current use, cpd, quitting, treatment, relapse, cancer). Data reported in the 1998 Surgeon General Report, Tobacco Use Among US Racial/Ethnic Minority Groups, suggests that there are unequal tobacco-related health indicators among racial and ethnic groups. Some differences are small, while others are relatively large and unexplainable for Latinos/Hispanics, blacks/African Americans, American Indian and Alaska Natives, and Asian Americans and Pacific Islanders. Because of small sample sizes, most data on minority racial/ethnic groups are reported in aggregate form, but some within-group data have been documented.
Latinos/Hispanics and tobacco disease continuum indicators
Latinos/Hispanics are the largest minority racial/ethnic group in the United States. In 2000, 12.5% of US citizens reported being Latino/Hispanic of any race . Of these, 58.5% reported Mexican, 4.8% Central American, 3.8% South American, 9.6% Puerto Rican, 3.5% Cuban, 2.2% Dominican and 17.3% other Latino/Hispanic . Latinos/Hispanics have the third highest poverty rate (21.9%) (see Fig. 2)  and third lowest median income of all the racial and ethnic groups ($34 241) . In 2003, only 57.0% of Latinos/Hispanics graduated from high school  (see Fig. 3). Of those in the labor force, 18.0% reported management/professional; 21.0% service; 23.1% sales/office; 2.7% farming, fishing and forestry; 9.6% construction, extraction and maintenance; and 13.2% production, transportation and material moving occupations  (see Fig. 4).
Initiation and ETS
Smoking initiation and ETS exposure among minority racial and ethnic groups are poorly understood. Regular onset of smoking occurs primarily before age 18 and Latinos/Hispanics have earlier onset than Asians/Pacific Islanders and blacks/African Americans [11,23,24]. The YRBS data indicate that age of initiation among Latinos/Hispanics is higher but similar to that of whites . Latinos/Hispanics are more likely to have ever smoked a whole cigarette compared to black/African American students and initiation diverges from black/African American students at ages 11–12 .
ETS exposure is low among Latinos/Hispanics in California  and is lower in Latino/Hispanic homes compared to other racial/ethnic groups except Somali immigrants . One study among children found that ETS exposure varies among ethnic groups; 10.3% of Mexican, 35.5% of Puerto Rican, 10.8% of Dominican and 16.4% of other Latinos/Hispanics were exposed to ETS compared to 45.8% of blacks/African Americans and 12.7% of whites .
MTF 30-day prevalence trends among 8th, 10th and 12th graders indicate that Latino/Hispanic smoking has declined since 2000 , but increased slightly from 2003 to 2005. Data from the 2003 YRBS indicate that 18.4% of Latino/Hispanic high school students smoked , but in 2005, 22% smoked . Disaggregate and pooled data reported from 1999 to 2001 NSDUH indicate that 10.8% of Latino/Hispanic youth ages 12–17 smoked, and among ethnic groups, 11% of Mexicans, 10.8% of Puerto Ricans, 9.6% of Central or South Americans and 12.4% of Cubans smoked . Combined data from the 1996–2000 MTF indicate that 30-day prevalence rates among 12th graders were 25.7% among Mexicans, 30.1% among Cubans, 29.6% among Puerto Ricans and 25.2% among other Latin American youth .
Current smoking has declined among adult Latinos/Hispanics since the 1970s . Data from the 2004 NHIS indicate that 15% of Latino/Hispanic adults smoked; this rate is the second lowest of all racial and ethnic groups (see Table 3) . According to the aggregate data reported using the 1999–2001 NSDUH, 23.1% of Latino/Hispanic adults smoked. The disaggregate data indicate that 23% of Mexican adults, 30.4% of Puerto Ricans, 21.3% of Central or South Americans and 19.2% of Cubans smoked . Other data also report that current smoking rates are highest among the Puerto Rican ethnic group [1,31].
Table 3. Percentage of people ≥ 18 years of age who were current smokers, by race/ethnicity, poverty and education, United States, National Health Interview Survey, 1994–2004.*
Tables adapted from National Health Interview Survey data reported in  accessed at http://www.cdc.gov/tobacco/data_statistics/tables/adult/table_2.htm. The NHIS in 1997 was redesigned and trend analysis and comparison with data years before 1997 should be conducted with caution. *Current smokers include those who smoked ≥ 100 cigarettes per day and who smoked every day or some days.†Data not collected in 1996.‡All racial ethnic groups are non-Hispanics except those categorized as Hispanic.§In 1997, the Office of Management and Budget changed its data collection guidelines to require that Native Hawaiian and Other Pacific Islander data be collected separately from Asian.¶Data on American Indians/Alaska Natives are small and data for a single year may be unstable or unreliable.**Additional categories were added to education in 1997.
Latinos/Hispanics in general tend to be light smokers. Only 4.6% of Mexican, 11.3% of Cuban, 11.5% of Puerto Rican and 4.8% of other Latin American youth in the 12th grade smoked a half-pack or more per day . The percentage of light (< 15 cpd) adult smokers has, in general, increased since the mid-1970s . At least 65% of Latino/Hispanic smokers smoked < 15 cpd . Data from the Multiethnic Cohort Study indicate that 59.0% of Latino/Hispanic men smoked ≤ 10 cpd and 27.8%, 11–20 cpd. Among women, 74.9% smoked ≤ 10 cpd and 18.4%, 11–20 cpd .
Quitting, relapse and treatment
Very little is known about quitting behaviors among Latino/Hispanic youth or adults. Data from the 2005 YRBS indicate that 53.4% of Latino/Hispanic youth tried to quit smoking . Quitting smoking has increased among Latino/Hispanic adults . Data from the 2000 NHIS indicate that 61.9% of Latinos/Hispanics want to quit and 42.9% who have ever smoked quit . According to the 1998 Health Outcomes Survey (HOS), 65.4% of Latinos/Hispanics in Medicare managed care indicate that they received advice to quit from a doctor or health care provider in the last 12 months, a rate lower than whites and blacks/African Americans .
The current data indicate that Latinos/Hispanics as an aggregate group suffer few disparities in tobacco-related cancers compared to other racial/ethnic groups. However, Latino/Hispanic women have the highest incidence rates of cervical cancer . Latino/Hispanic men and women have the lowest incidence and death rates of lung cancer, and lung cancer incidence rates decreased for Latino/Hispanic women along with American Indians/Alaska Natives, while these rates remained stable for other women. Rates of lung cancer have declined among Latino/Hispanic men . The incidence and death rates of oral/pharynx cancer were lowest among Latino/Hispanic men compared to other men (see Tables 1 and 2).
Blacks/African Americans and tobacco disease continuum indicators
In the 2000 Census, 12.2% of Americans reported being black or African American as their only race/ethnicity . Four per cent of blacks/African Americans reported being foreign-born, most whom originate from the Caribbean and Africa . Compared to other racial and ethnic groups, blacks/African Americans have the highest poverty rates (24.7%) (see Fig. 2), the lowest median income ($30 134)  and 80.0% completed a high school education  (see Fig. 3). Of those in the labor force, 25.2% reported management/professional; 22.0% service; 27.3% sales/office; 0.4% farming, fishing and forestry; 6.5% construction, extraction and maintenance; and 18.6% production, transportation and material moving occupations  (see Fig. 4).
Initiation rates and ETS
Smoking initiation rates have declined among blacks/African Americans  and delayed onset has been documented consistently as a phenomenon among black/African American smokers. Blacks/African Americans begin regular smoking primarily after the age of 18 [11,23,36–42], and overall have lower smoking initiation rates compared to whites  and Latinos/Hispanics during adolescence . Data from YRBS show similarities in initiation rates at ages 9–10 for blacks/African Americans, whites and Latinos/Hispanics, but at ages 11–12, black/African American initiation rates were substantially lower and diverged from whites and Latinos/Hispanics . One study indicated that black/African American youth had higher initiation rates than whites, but lower rates of regular use compared to whites and Latinos/Hispanics . Existing data do not indicate within group differences in smoking onset; the mean age of smoking onset among Caribbean-born blacks was 18.6 years, Haitian-born, 19.5 years and US-born blacks 18.3 years .
Some studies suggest that ETS exposure among black/African American children is greater than all other ethnic groups [26,27]. Studies have found higher levels of serum cotinine and hair cotinine in black/African American children compared to white children, despite lower levels of reported ETS exposure [45,46]. Higher levels of serum cotinine were found among black/African American children compared to Latino/Hispanic children after controlling for ETS exposure . Thirty-seven per cent of black/African American adults compared to 37.4% of whites and 35.1% of Mexican Americans reported exposure to ETS .
Data from MTF indicate that 30-day prevalence rates among youth have decreased steadily since 1998 . Trends in 30-day prevalence indicate that rates overall have been consistently and substantially lower among black/African American youth compared to other racial/ethnic groups [1,9,36,43,49]. The 2005 YRBS indicate that 12.9% of black/African American high school students smoked . The 1999–2001 NSDUH data indicate that 7% of black/African American youth smoked, which was lower than all racial/ethnic groups . Combined data from the 1996–2000 MTF indicate that 14.3% of black/African American 12th graders smoked .
Among adults, smoking rates have also decreased. In 1965, rates were 45.8%  and in the 2004 NHIS, 20.2% of black/African American adults reported smoking, with rates slightly lower than the rate among whites (22.2%) (see Table 3) . Current smoking rates among black/African American men were higher than white men in 1965 but rates have declined among black men and women since the mid-1970s . In 2004, 23.9% of black/African American men and 17.2% of black/African American women smoked . Haitian-born and Caribbean-born blacks report lower rates of current smoking than American-born blacks [42,51]; this study found that 7.1% of Caribbean-born, 9.7% of Haitian-born, and 23.4% of US-born blacks smoked . Trend data that delineate smoking rates for foreign-born versus native-born blacks have not been published.
Similar to Latinos/Hispanics, blacks/African Americans tend to be light smokers. Only 2.6% of 12th grade black/African American youth smoked a half-pack or more per day . Data from YRBS also indicate that blacks/African Americans and Latinos/Hispanics smoked fewer cpd than white students . The percentage of black/African American adults who smoked < 15 cpd has increased since the 1970s [1,52]. More than 60.0% of black/African American adults smoked < 15 cpd and smoked fewer cpd than whites . Less than 10.0% of blacks/African Americans smoked ≥ 25 cpd . Data from the Multiethnic Cohort Study indicate that 42.9% of black/African American men smoked ≤ 10 cpd and 40.0%, 11–20 cpd, and among women, 59.7% smoked ≤ 10 cpd and 30.6%, 11–20 cpd . Despite the fact that blacks/African Americans in general are light smokers, black/African American adolescents show higher cotinine-to-cigarette per day ratios than white youth  and black/African American adults also have higher serum cotinine than white [55–58] and Mexican American adult smokers, who smoke the same number of cpd [1,56,58].
Quitting, relapse and treatment
Little is know about quitting among black/African American youth; however, data from the 2005 YRBS indicate that 61.8% of black/African American youth tried to quit . Quit rates have increased since 1978 among black/African American adults, but were lower than whites and other racial and ethnic groups [52,59,60]. In the 2000 NHIS, 68.4% of blacks/African Americans desired to quit, but only 37.3% who have ever smoked quit. This quit rate was lower than that of any other racial or ethnic group . Data from the Multiethnic Cohort Study indicate that rates of quitting were significantly lower among black/African American compared to white men (42.2% versus 55.5%), but were not significantly different from Japanese, Native Hawaiian or Latino/Hispanic men . Some data suggest that there are few differences in mean quit attempts among blacks/African Americans, whites and Latinos/Hispanics . In a 1998 survey of Medicare managed care enrolled smokers, 67.8% of blacks/African Americans reported receiving advice to quit , and studies have found that younger blacks/African Americans were less likely than whites to receive advice to quit .
Blacks/African Americans have the highest overall incidence and death rates of all cancers (see Tables 1 and 2). Although rates are declining, black/African American men have the highest lung cancer incidence and death rates and there is an enormous disparity between black/African American men and Latino/Hispanic men, who have the lowest incidence and death rates. Black/African American men also have a younger age of lung cancer diagnosis than other groups . Deaths from stomach cancer are more than 200% higher and for esophageal and oral cavity cancer, 75% higher among black/African American compared to white men . Lung cancer incidence and death rates are lower among black/African American women compared to black/African American men, but black/African American women have the highest incidence rates of lung cancer and the second highest death rates after white women . Lung cancer death rates have not declined for black/African American women .
American Indians, Alaska Natives and tobacco disease continuum indicators
In 2000, American Indians and Alaska Natives, not including those who report one or more race, comprised 0.9% of the US population . Over 500 federally recognized tribes, villages and communities exist in the United States. The Cherokee community is the largest, with about 302 569 members , but many tribes have less than 1000 members . Based on 3-year averages of aggregate data, data indicate that 24.3% of American Indians/Alaska Natives live in poverty and they have the second lowest median income ($33 132) of all racial/ethnic groups . Although poverty rates vary among tribes, rates of poverty among all American Indian/Alaska Natives exceed that of whites and Asian Americans . Seventy-one per cent have at least a high school education, which varies by tribe . Of American Indians/Alaska Natives in the labor force, 24.3% reported management/professional; 20.6% service; 24.0% sales/office; 1.3% farming, fishing and forestry; 12.9% construction, extraction and maintenance; and 16.8% production, transportation and material moving occupations  (see Fig. 4). Some studies report data on American Indians separately from Alaska Natives and few report regional or tribal differences. We report differences where possible and use the label American Indians/Alaska Natives to represent the aggregate category reported in surveys.
Initiation and ETS
Although smokeless tobacco use is not the focus of this paper, it is worth noting that smokeless tobacco use rates range from 18 to 60% among youth and that initiation can begin as early as pre-school and kindergarten years among Alaska Natives [64–66]. Studies also report that 68.9% of girls and 79.1% of boys at Indian Health Service sites experiment with smokeless tobacco . Among American Indians, smoking initiation rates have been reported in middle school . In addition, the use of Ikmik, a smokeless form of tobacco, is used at all ages from youth to adulthood .
Very little is known about ETS exposure among American Indians and Alaska Natives. Studies indicate that in a combined category of Native American and South-east Asian children, ETS exposure was higher than Latino/Hispanic and Somali immigrants but lower than Black/African American children . The US Women's Determinants Study found that home ETS exposure is higher among American Indians/Alaska Natives than other racial/ethnic groups .
American Indian/Alaska Native youth and adults have the highest rates of tobacco use in the United States . Although many studies indicate that ceremonial use of tobacco is common among American Indians, recent studies indicate that non-traditional use of tobacco occurs frequently among American Indians . Recent data from the YRBS and MTF are not published for American Indian/Alaska Native youth and very little trend data are available. From 1990 to 1994, the current smoking rate among 12th grade American Indians/Alaska Natives was 41.1%, but from 1995 to 1999 rates were 42.3% . Data from the 1999–2001 NSDUH indicate that 27.9% of American Indian/Alaska Native high school students were current smokers . Combined data from the 1996–2000 MTF indicate that 46.1% of American Indian 12th graders smoked .
Prevalence rates among adults have not declined consistently since the 1970s . According to the 2004 NHIS, 33.4% of American Indian/Alaska Native adults smoked (see Table 3) . Rates are higher among men compared to women (37.3% versus 28.5%) . Higher rates have been reported in Alaska and the Northern Plains and lower rates reported in the South-west [1,72]. Forty-four per cent of those in the Northern Plains, 39% in Alaska, 31.9% in the East, 30.9% in the Pacific Coast and 21.2% in the South-west smoked .
The 1997 BRFSS data from Alaska indicate that 41.3% American Indians/Alaska Natives smoked . Alaska Native adult current use rates range from 39 to 50% and among women, 36–68% . Pregnant Alaska Native women have smoking prevalence rates of 39–44% . Alaska women in rural areas have higher rates of tobacco use than women in urban areas . Urban American Indians report higher smoking rates than rural American Indians .
Seventeen per cent of American Indian 12th graders smoked a half-pack or more per day; this rate was higher than that among white 12th graders (15.9%) . The percentage of light smokers (< 15 cpd) has increased among adults . About 50% of American Indians/Alaska Natives smoked < 15 cpd  and men were more likely than women to be heavy smokers (≥ 25 cpd) . Heavy smoking among American Indians/Alaska Natives, however, was lower than white adults . The number of cpd varies by region, but in all regions, light smoking exceeded 70.9% . American Indian adults report smoking on average 10 cpd .
Quitting, relapse and treatment
Very little is known about quitting among American Indians and Alaska Natives. Quit rates have fluctuated since the 1970s . In the 2000 NHIS, 69.8% reported a desire to quit and 40.9% of American Indians/Alaska Natives who ever smoked quit . In the 1998 survey among Medicare managed care smokers, 72.2% of American Indians/Alaska Natives reported that they had received advice to quit, a rate equivalent to that of whites, who report the highest rates of receiving advice to quit .
American Indian/Alaska Native men and women have the second lowest lung cancer incidence rates of all racial/ethnic groups and the third highest lung cancer death rates (see Tables 1 and 2). These same data indicate that lung cancer incidence rates decreased for American Indian/Alaska Native women, but lung cancer deaths did not decline among American Indian/Alaska Native men . No declines were observed in cancers of the stomach and larynx among American Indians/Alaska Natives .
Lung cancer death rates reported by Indian Health Service (IHS) areas parallel smoking rates in those areas . Data on lung cancer mortality from 1994 to 1998 indicate that American Indians/Alaska Natives in Alaska (78.1) and in the Northern Plains (96.9) have higher rates of lung cancer compared to the total US population of all racial/ethnic groups (57.8); rates were substantially lower among American Indians/Alaska Natives in the East (37.0), the Pacific Coast (39.5) and the South-west (14.1) . Stomach cancer mortality from 1994 to 1998 was significantly higher in Alaska (16.3), the Northern Plains (8.9) and the South-west (10.3) compared to the total population of all racial/ethnic groups (5.1) .
Another study indicated that Alaska Natives have higher tobacco-related cancers including lung, oral/pharynx, stomach, kidney, cervix and pancreatic cancers than whites . Lung cancer 5-year survival rates were 5% lower among Alaska Natives compared to whites from 1984 to 1999 and also lower for esophageal, pancreatic, stomach, oral, pharynx, kidney and cervical cancers .
Asian Americans, Hawaiians/Pacific Islanders and tobacco disease continuum indicators
In 2000, Asian Americans comprised 3.6% and Native Hawaiians/Pacific Islanders 0.1% of the US population . Asians and Pacific Islanders consist of Asians born in the continental United States; Asian immigrants and refugees; and natives of Hawaii, Guam, American Samoa, Federated States of Micronesia, Republic of the Marshall Islands, Republic of Palau and the Commonwealth of the Northern Miriana Islands. Hawaiians make up the largest Pacific Islander group  and Chinese the largest Asian American group in the United States . The aggregate group Asian American/Pacific Islander consists of more than 32 ethnic and language groups. Over 500 languages and dialects are represented among the aggregate group . Asian Americans have the lowest poverty rate of all racial and ethnic groups (9.8%) (see Fig. 2) and the highest median income ($57 518). Native Hawaiians and Pacific Islanders have the second highest median income ($56 664) and a poverty rate of 13.2%. Because of small sample sizes, the Native Hawaiian and Pacific Islander income estimates are based on 3-year averages . Eighty-eight per cent of Asian Americans and  82.0% of Hawaiians and Pacific Islanders have graduated from high school  (see Fig. 3). Of Asian Americans in the labor force, 44.6% reported management/professional; 14.1% service; 24.0% sales/office; 0.3% farming, fishing and forestry; 3.6% construction, extraction and maintenance; and 13.4% production, transportation and material moving occupations  (see Fig. 4). Of Hawaiians and Pacific Islanders in the labor force, 23.3% reported management/professional; 20.8% service; 28.8% sales/office; 0.9% farming, fishing and forestry; 9.6% construction, extraction and maintenance; and 16.5% production, transportation and material moving occupations . We report ethnic differences where possible.
Initiation and ETS
Little is known about differences in initiation rates among Asian ethnic groups, but existing data suggest that there are within-group differences. Asian American youth are less likely to begin smoking in early adolescence [80,81]. Data from the 2000 NYTS indicate that smoking initiation rates were much lower among middle school Asian American (39.5%) compared to Hawaiians/Pacific Islanders (44.0%), blacks/African Americans (47.6%), Latinos/Hispanics (45.8%) and whites (47.2%). However, among high school students, smoking initiation was highest among Asian American youth (42.5), followed by Latinos/Hispanics (33.9%), whites (33.2%), blacks/African Americans (31.5%) and Hawaiians/Pacific Islanders (24.9%) . Chinese and Koreans initiate later in life than whites  and Asian Americans/Pacific Islanders were more likely to start smoking regularly after the age of 18 . Little is available on ETS exposure, but one study indicated that 31% of Vietnamese women were exposed daily to ETS .
Recent trend data from the YRBS and MTF have not been published for Asian American/Pacific Islander youth. Depending on the survey, current smoking rates vary among Asian Americans/Pacific Islanders. Data from the 1999–2001 NSDUH indicate that 8.1% of Asian American youth and 11.0% of Hawaiian/Pacific Islander youth smoked . For Asian ethnic groups, current smoking was 10.6% among Koreans, 8.7% among Asian Indian, 7.4% among Filipino, 5.8% among Chinese, 5.2% among Japanese and 6.8% among Vietnamese youth . Combined data from the 1996–2000 MTF indicate that 20.4% of Asian American 12th graders smoked . Data from the 2000 NYTS indicate that 32% of Hawaiian and Pacific Islander high school students smoked . Smoking rates have been reported as high as 38% among Vietnamese male high school students . Another study reported that 8.6% of Filipino 8.3% of Korean, 7.4% of Japanese, 7.2% other Asian American and 2.8% of Chinese youth in California smoked .
Smoking rates decreased among Asian American/Pacific Islander adults from 2000 to 2004 (see Table 3). According to the 2004 NHIS, Asian American adults report lower current smoking than all racial and ethnic groups (11.3%) and rates were particularly low among Asian American women (4.8%) . Data from the 1999–2001 NSDUH indicate that 16.2% of Asian Americans adults smoked . Among ethnic groups, 12.3% of Chinese, 12.6% of Asian Indian, 14.8% of Filipino, 19.0% of Japanese, 26.5% of Vietnamese and 27.2% of Korean adults currently smoked . Asian immigrants from South-east Asia (Vietnam, Cambodia and Laos) have reported rates ranging from 34 to 43% . Current rates among Vietnamese men in California have been reported from 35 to 57% . Korean men in California report high smoking rates (38.7%); however, Korean women report low rates (6.0%) . The 1997 BRFSS data indicate that 17.3% of Asians/Pacific Islanders in Hawaii smoked . Combined data among Native Hawaiians indicate that 28.9% of men and 27.6% of women smoked .
Few data have been published on the number of cigarettes smoked per day among Asian Americans/Pacific Islanders. Only 4.9% of Asian American 12th graders smoked a half-pack or more per day . The percentage of adult light smokers (< 15 cpd) has increased . More than 70% of Asians/Pacific Islanders smoked < 15 cpd . Data from the Multiethnic Cohort indicate that 22% of Japanese American men smoked ≤ 10 cpd and 43.6%, 11–20 cpd, and among Native Hawaiians, 25.2% smoked ≤ 10 cpd and 39.7%, 11–20 cpd . Among women, 53% of Japanese Americans smoked ≤ 10 cpd and 34.9%, 11–20 cpd, and among Native Hawaiians, 44.3% smoked ≤ 10 cpd and 37.1%, 11–20 cpd . One study among smokers showed slower clearance of and reduced intake of nicotine among Chinese Americans compared to whites .
Quitting, treatment and relapse
Quit rates have increased since the 1970s . In 2000 NHIS, 67.8% of Asian Americans desired to quit and 44.7% who have ever smoked quit . In the 1998 survey of Medicare managed care enrolled smokers, 54% of Asian Americans/Pacific Islanders reported receiving advice to quit . Data on Hawaiians indicate that 35.6% of men and 27.6% of women quit smoking . Compared to Cambodian men, Laotian and Vietnamese men have lower rates of quitting . Little is known about relapse behavior among Asian Americans/Pacific Islanders.
Asian American/Pacific Islander men and women have the third lowest lung/bronchus cancer incidence and the second lowest death rates (see Tables 1 and 2). Asian Americans/Pacific Islanders have the highest incidence rates of stomach cancer . Incidence rates for oral cancer decreased among all men except Asian Americans/Pacific Islanders . Data from the California Multiethnic Cohort Study indicate that Native Hawaiian (263.9) lung cancer incidence rates were higher than whites (158.3), Japanese Americans (121.4) and Latinos/Hispanics (79.2), but equivalent to blacks/African Americans (263.9) . Data from SEER 1988–99 indicate that the incidence and death rates of cancer among Hawaiians is lower than that of black/African American and white males .
Since the 1998 Surgeon General Report, Tobacco Use Among US Racial/Ethnic Minority Groups was published, new data have been published and provide additional insight on the tobacco disease continuum indicators. Much of the data combine prevalence, incidence and mortality rates across years to develop stable estimates for minority groups and data from states such as California help to increase our understanding of within ethnic group differences. However, gaps in knowledge still remain and inhibit our ability to understand the link between smoking and tobacco-related cancer outcomes. Additional information is needed on smoking initiation, ETS exposure, quitting and relapse patterns among minority racial/ethnic groups with a specific focus on within-group differences.
Other factors may also help to explain existing disparities. Poverty, education and occupation may impact TRHD independently and are likely to interact with race and ethnicity to impact disparities  at various points along the tobacco disease continuum. Race and ethnicity play important roles in understanding how minorities participate in systems which affect their poverty, levels of education and where they work . Data on these interactions and stratified data are lacking and difficult to obtain because of small sample sizes.
Poverty and tobacco disease continuum indicators
Poverty is defined as a measure of family income less than the Census Bureau's official family money income threshold . Family income thresholds vary by family size, but the official poverty thresholds are not adjusted to geographic location . Cumulative adverse health effects result from living in poverty . Population attributable death rates due to poverty increased between the 1970s and 1990s, especially among blacks/African Americans . Poverty is an important variable to examine in tobacco-related health disparities because it is associated with multiple measures along the tobacco continuum; rates of poverty are more than twice as high among blacks/African Americans, American Indians/Alaska Natives and Latinos/Hispanics than whites, Asian Americans and Pacific Islanders; and the condition of poverty coupled with tobacco use might lead to multiple and cumulative disparities across the life-span.
Initiation and ETS
Low socio-economic status has predicted smoking initiation in longitudinal studies . Studies have found that children in low-income areas are more likely to begin smoking than those in higher-income areas . Children from ethnically diverse low-income and high-poverty neighborhoods are exposed to ETS , and exposure is as high as 89% among low SES children whose mothers' smoked . Furthermore, ETS exposure among adults has been associated inversely with income . However, some studies indicate that ETS exposure is low in low-income housing .
Individuals below poverty have consistently had higher rates of smoking than those at or above poverty and the disparities gap has not changed in the last 10 years (see Table 3). In 2004, 29.1% of adults below poverty were current smokers, with rates slightly higher among men (31.9%) compared to women (27.1%) .
Quitting, relapse and treatment
Individuals below poverty are less likely to quit smoking . In 2000, 33.6% of ever smokers who were below poverty quit smoking compared to 49.9% of those at or above poverty . Pregnant women below poverty were less likely to report a quit attempt . Lower rates of quitting may be explained by unequal access to valid treatment . Only half of Medicaid recipients in 1998 were covered for one or more tobacco dependence treatment  and few studies have investigated the effectiveness of cessation treatments for the poor .
Disparities exist in lung cancer incidence and mortality among poor men and women. SEER trends indicate that the lung cancer incidence rate for men in high-poverty counties was at least 12% greater than men in low-poverty counties from 1975 to 1999. Lung cancer incidence from 1988 to 1992 increased with increasing census tract poverty rates for white and black/African American men and women and Asian American/Pacific Islander men. On the contrary, lung cancer incidence rates were higher in low-poverty census tracts than in high-poverty census tracts for Latino/Hispanic men and women and American Indian/Alaska Native women. Trends in female lung cancer indicate that the incidence rate in high poverty counties was 11% greater than the rates for women in counties with poverty levels between 10 and 20% . Compared to men in low-poverty counties, the lung cancer mortality for men in high-poverty counties was 7% greater in 1975 and 25% greater in 1999 . Differences in lung cancer mortality among women diminished among area poverty groups from 1975 to the 1990s, and differences were no longer significant in 1999. Trends from 1975 to 1999 indicate that men in high-poverty areas experienced at least 18% higher lung cancer mortality than men in low-poverty areas .
Education and tobacco disease continuum indicators
Education, like poverty, has also been associated with health outcomes. Studies indicate widening gaps in mortality by education [104,105]. High levels of education, in general, are associated with lower risk of ill health and deaths due to cancer and cardiovascular disease across multiple cultures  and the life-span . Cigarette smoking is associated with lower education and studies show differences in smoking by educational status.
Initiation and ETS
Education has been associated independently with initiation and exposure [52,53]. Studies show that children of parents with low education were more likely to try smoking [108–110]. One European study found that among women, higher education was associated with a higher relative risk of smoking initiation . Low education has also been associated significantly with ETS exposure among young females  and ETS inversely associated with education among adults . One study among women 40 and over found that higher ETS exposure was associated with lower education .
Thirty-day prevalence rates among children of parents with low education have decreased since 1999 . Children of parents with low education are more likely to be current smokers [9,113–116], with the exception of children of parents with the lowest levels of education. Among 12th graders, 19.8% of those with the lowest parental education smoke compared to 18% of those with the highest parental education . Among adults, smoking is more prevalent among the lower educated [23,112] and smoking rates have not been shown to have a direct inverse relationship to educational attainment [4,9]. Current smoking rates are highest among those with a General Education Development (GED) (39.6%) and lowest among those with a graduate degree (8%) (see Table 3). Data from the 2004 NSDUH indicate that among adults, 34.8% of those who had not completed a high school education, 30.4% of high school graduates who did not attend college, 29% of people with some college and 13.6% of college graduates smoked . Full-time college students are less likely to smoke than 18–22-year-olds not enrolled full-time in college .
Little is known about the number of cpd among educational groups. Those with higher levels of education smoke fewer cpd . One study found that as maternal education increased, the number of cigarettes smoked by adolescent offspring decreased .
Quitting, relapse and treatment
No direct inverse relationship exists between quitting smoking and educational attainment, but overall those with less education are less likely to quit successfully [52,59,119,120]. In the 2000 NHIS, ever quit rates were lowest among those with a GED certificate (33.6%), 55.8% among those with < 8 years of education, 53.7% among those with an associate degree and highest among those with a graduate degree (74.4%). Those with higher education are more likely to report higher mean quit attempts than lower education groups . Although few studies have examined smoking relapse among the lower-educated, one study among lung cancer patients found that low education predicted current smoking and shorter time to relapse . In general, advice to quit is lower among the less educated .
Cancer rates are generally higher among those with low education, and lung cancer risk  and mortality is also higher among those with low education compared to high education groups . Lung cancer mortality of men with elementary education or less was 32% higher than better educated men . Another study also found a strong inverse education gradient for lung cancer . Data from the California Multiethnic Cohort indicate that people attending some college have a decreased risk of lung cancer .
Occupation and tobacco disease continuum indicators
In addition to poverty and education, tobacco use coupled with occupational exposure to certain chemicals, dust, pesticides, radiation and industrial processes increases the risk for tobacco related diseases [6,125]. Nearly 65% of US citizens participate in the labor force . A larger proportion of Asian Americans and whites are employed in management/professional jobs compared to blacks/African Americans, American Indians/Alaska Natives, Hispanics/Latinos and Hawaiians/Pacific Islanders, many of whom are employed in service, sales and office positions. Differences in occupations may impact racial and ethnic differences in tobacco-related disparities.
Initiation and ETS
The work-place is a major source of environmental tobacco smoke exposure and about 40 000 deaths are due to environmental tobacco smoke each year . ETS can decrease the productivity of the worker and cause employers to face liability for non-smoker's health . Furthermore, ETS exposure among adults has been associated inversely with occupational class . Blue-collar and service workers are significantly less likely to be protected by smoke-free policies than white-collar workers [128–130]. Bartenders and waiters/waitresses are less likely to be covered by a smoke-free policy and are more likely to be exposed to ETS even when covered by such a policy . Migrant, seasonal and other workers who are exposed to tobacco leaves may suffer from green tobacco sickness, an occupational illness resulting from transdermal nicotine exposure . The work-place may also contribute to smoking initiation. One-third of youth begin smoking regularly at work  and those in blue-collar and service position are more likely to initiate smoking earlier in life than white-collar workers [6,133,134].
One in three blue-collar and service workers smoke compared to one in five white-collar workers  and, in general, smoking rates are higher among blue-collar workers [6,133,136]. Rates vary within the blue-collar occupation . In specific trades, such as construction workers, the prevalence can be as high as 40% .
Blue-collar workers (27.5%) and farm workers (27.0%) were more likely to smoke ≥ 25 cpd and estimates were significantly lower among white-collar (18.0%) and service workers (16.7%) . Eight per cent of adolescent workers in service positions reported smoking > 10 cpd .
Quitting, relapse and treatment
Blue-collar and service workers are not as successful in quitting smoking as white-collar workers [6,138] even though they make the same number of quit attempts . Fifty-one per cent of white-collar, 36.8% of blue-collar, 41.4% of farm workers and 32.8% of service workers quit successfully .
Occupational exposures place workers at high risk for lung cancer, and some of these occupational agents are synergistic with smoking, increasing lung cancer risks [6,125,139] and diseases such as restrictive and chronic obstructive lung disease . Studies in multiple countries indicate an occupational class gradient in lung cancer mortality among men .
In summary, those in poverty, with lower education and in service and blue-collar occupations generally fare worse across the tobacco disease continuum. Those with a GED certificate have higher smoking and lower quit rates than all educational levels. The effects of poverty on cancer vary by gender and racial ethnic group and among Latinos/Hispanics, those in the lowest poverty census tracts have lower cancer incidence than those in the highest poverty census tracts. Data on the number of cigarettes smoked per day by poverty level and education are lacking. While social factors may help to explain differences observed among racial/ethnic groups, focusing only upon ‘disparities’ in health is problematic because it is sanitized from any connection to structural inequality or other social contexts as producers of inequalities in health . The question remains as to how to eliminate TRHD with the data that we have. Examining work-force diversity, inequities in research and inequalities in health care may help resolve challenges in eliminating disparities.
‘Diversity’ describes the integration and inclusion of races, ethnicities, genders and groups from different geographies, cultures and social classes into organizations, decision-making tiers, institutions and systems from which they are and have historically been excluded. Diversity creates a climate where there is variety in the quantity and quality of interactions  and equitable power brokering among stakeholders at multiple levels. Diversity may create research environments where the composition and group exchanges influence the generation and validation of research ideas, philosophy, conceptualization, theoretical approaches, design, implementation and interpretation. When health care organizations are diversified, health providers better understand medical conditions and treatments , and in classroom settings, diversity results in greater gains in learning and citizen engagement . Students from diverse backgrounds in training settings may help learners to challenge assumptions and broaden perspectives regarding racial, ethnic and cultural differences [142,144]. The integration and inclusion of various groups into different environments, structures and decision-making bodies may alter scientific thinking and approaches to tobacco health disparities research.
Diversity may mediate health outcomes [15,145] and is associated with improved access to care for minority patients and greater patient choice and satisfaction . When provided with a choice, minority patients are more likely to select health care professionals of their own race or ethnicity  and are generally more satisfied with the care they receive from minority professionals [146,147]. Language-concordant physicians contribute to improved patient communications and healing . Furthermore, racial and ethnic minority health care providers  and minority psychologists  are significantly more likely than their white peers to serve and practice in minority and medically underserved communities. Cantor et al.  found that minority and women physicians and those from lower socio-economic backgrounds are more likely to serve minority, low-income and Medicaid populations—those who often suffer from disparities.
Contact with minority patients may improve our understanding of the causes and consequences of racial/ethnic health disparities. Health providers and researchers from diverse backgrounds bring cultural perspectives and experiences to research and provider teams, which can increase the likelihood that socio-cultural issues influencing health outcomes will be addressed in research design and inquiry . For example, some researchers have a wealth of knowledge on the relevance of cultural norms and experiences with tobacco, historical experience with mainstream cultures and researchers, healing practices, resiliencies and strengths, community social norms and historical relationships with tobacco (e.g. working in tobacco companies and on farms, receiving funding for social and cultural events). Health professionals with this knowledge may help to increase the acceptability and value of community experiences and practices. These professionals may also increase the acceptability of community-based participatory research and help to increase enrollment of minority patients into clinical trials  and intervention studies that seek to reduce or prevent tobacco use or exposure and improve time to diagnosis and treatment of tobacco-related cancers.
Attending to diversity in research and health organization practice may help us to address related inequities in research and intervention practice. ‘Inequities’ represent injustice and unfairness  in scientific practice and knowledge allocation. Inequities in research include inequities in scientific inquiry; validation and value for research and unique methodologies in small and traditionally underserved and understudied populations; the social integration of a health disparities agenda into mainstream tobacco control research; the application of research to populations affected by tobacco disparities; representation of underserved groups in research; the translation and delivery of effective interventions and services; and the development of applicable and relevant treatments for tobacco-related conditions.
Research priorities in non-mainstream cultures differ from priorities in mainstream cultures , which may contribute to an under-representation of tobacco-related research questions relevant to minorities. For example, in general, why do Alaska Native/American Indian youth and adults have the highest rates of tobacco use but lower rates of tobacco-related cancers compared to blacks/African Americans? While numerous community and institutional factors may impact the capacity to explore this question, understudied research may be impacted by perceived bias in the peer review system toward grants addressing disparities , the availability of funding mechanisms, interest of journals and acceptance of relevant peer-reviewed publications, and the value and reward within academic institutions and other organizations for tobacco-related disparities research.
Minority groups are typically under-represented in research . For example, the National Health Interview Survey which reports annual tobacco prevalence in the United States does not collect data on US territories, Hawaiians and Pacific Islanders . Under-representation inhibits the ability to analyze and detect between- and within-race/ethnic group differences and conduct multi-level analyses that incorporate social variables such as poverty, education and occupation. Minority groups are also under-represented because standard sampling methods may not capture prevalence rates among small minority groups such as lesbian, gay, bisexual or transgender (LGBT) minorities. Researchers are often slow to accept or even test alternative sampling methods (i.e. respondent-driven sampling) that may hold promise . The lack of these data inhibits the development of prevention and cessation interventions for populations with the greatest risks for tobacco use (i.e. Alaska Natives, LGBT).
The gold standard, randomized controlled study designs, may not be feasible for some populations and the rejection of other designs suggests that there is a lack of sensitivity among grant reviewers for the context of research in racial/ethnic groups . For example, when communities are small and closely connected, the randomized control design may not be suitable. Cessation interventions among racial/ethnic groups are lacking and most are quasi-experimental designs . Furthermore, few recommended interventions (e.g. advice to quit, nicotine replacement therapy) have proven effectiveness in reducing disparities among racial/ethnic groups. Inequities in scientific practice and knowledge allocation impact our ability to systemically address disparities related to tobacco.
‘Inequalities’ are defined as the quality of being unequal or uneven . Inequalities lie in access to care, quality of health care, socio-economic indicators that impact health care, and psychosocial and environmental resources (see Fig. 1). With the exception of psychosocial resources, multiple US national surveys track many of these inequalities.
The National Health Care Disparities Report, 2005 indicates that disparities in access to and quality of care among minority racial/ethnic and poor groups still exist . For example, poor hospitalized smokers compared to high-income hospitalized smokers have increasing disparities related to receipt of advice to quit from health providers; this same inequality is observed among Latinos/Hispanics compared to non-Hispanic whites . Access to health insurance from 1999 to 2003 was much lower among blacks/African Americans and American Indians/Alaska Natives compared to whites; among poor compared to high-income people; and among people with high school education or less compared to those with some college education. Latinos/Hispanics of every income and education level are less likely than non-Hispanic whites to have health insurance . Even with access to care, inequalities still exist. Black/African American lung cancer patients are less likely to receive invasive staging for lung cancer, and even when there is staging, black lung cancer patients are less likely to receive recommendation for surgery . Overall, blacks/African Americans, Latinos/Hispanics and American Indians/Alaskan Natives have poorer quality of care and worse access to care compared to whites  and factors such as discrimination and social prejudices may contribute to the disparities among racial/ethnic groups.
This paper suggests that in addition to addressing disparities, there is a need to address diversity, inequities and inequalities in efforts to eliminate TRHD. Data from national, state and local studies indicate that disparities represent relatively large or small differences between racial/ethnic groups and vary along the tobacco disease continuum. Based on the dose–response model for lung cancer risk , one might assume that duration of regular smoking and number of cigarettes smoked per day might support disease outcomes among racial/ethnic groups. Available evidence on blacks/African Americans, American Indians and Alaska Natives are not consistent with the model. Socio-economic differences among the racial/ethnic groups may not explain these inconsistencies fully. Additional data, strategies and models are needed to resolve the disparities puzzle among some minority racial/ethnic groups.
We should observe similar if not higher cancer incidence and death rates among American Indians/Alaska Natives compared to blacks/African Americans. American Indians/Alaska Natives begin using tobacco earlier in life, current smoking rates are substantially higher, the percentage of light smoking is lower, but quit rates slightly higher compared to blacks/African Americans. Smoking prevalence rates have consistently and substantially declined among blacks/African Americans, but not among American Indians/Alaska Natives. Only a small percentage of blacks/African Americans and American Indians/Alaska Natives smoke heavily; whites have substantially higher rates of heavy smoking compared to other racial/ethnic groups . Socio-economic indicators for American Indians/Alaska Natives and blacks/African Americans are quite similar, yet we observe more devastating outcomes among blacks/African Americans.
Part of the challenge is that most American Indian/Alaska Native tobacco continuum indicators are reported in aggregate form, which impacts our ability to fully understand how consistent the epidemiological data are with the tobacco disease causal paradigm. Alaska Native cancer data indicate higher cancer incidence rates of six different tobacco-related cancers when compared to whites,  and regional differences in smoking parallel lung cancer rates among American Indians/Alaska Natives. Misclassification of American Indians/Alaska Natives in cancer registries may underestimate their cancer incidence and overestimate cancer among whites. One study found that 52.2% of cases in the Oregon, Washington and Idaho cancer data registries were misclassified; 92% were coded as white . None of these factors explain differences in cancer outcomes among American Indians/Alaska Natives compared to blacks/African Americans.
Asian Americans/Pacific Islanders and Latinos/Hispanics experience few tobacco-related cancer disparities. Adult smoking rates are lower in these groups compared to other racial/ethnic groups, and other indicators along the tobacco disease continuum are similar for Asian Americans/Pacific Islanders and Latinos/Hispanics. One might assume that as both aggregate groups have the lowest smoking rates, but substantially different socio-economic indicators, that Latinos/Hispanics might fare worse than Asian American/Pacific Islanders; but this is not the case. The higher lung cancer incidence rate for Latinos/Hispanics in low-poverty census tracts compared to high-poverty census tracts is puzzling. Focusing on variations in smoking among Puerto Ricans, Hawaiians, Pacific Islanders and South-east Asian men is needed to understand more clearly the dose–response model for these groups within their socio-economic context.
Occupation distribution, and therefore exposure to toxic chemicals, dust and fumes, may partially explain existing disparities . One study found racial differences in exposure to environmental and occupational hazards  and another found increased risk for lung cancer among black/African American male farm workers, auto mechanics, painting machine operators, furnace operators, garbage collectors and cutting machines operators in Michigan . Few studies have compared environmental risks for lung cancer and how work environment exposure may be experienced differentially among racial/ethnic groups. It is not known if occupational exposure to lung carcinogens varies when work-sites or work areas are predominately minority.
The experience, longevity and depth (e.g. 125%, 200%) of poverty may vary by race/ethnicity, geography, rural/urban, type of housing and neighborhood conditions, thereby resulting in differential effects that impact the tobacco disease continuum. Cumulative effects of economic deprivation may also be experienced differentially across the life-span [161,162]. For example, poverty is quite prevalent among the employed; 38.9% of those employed were below poverty . The effects of economic deprivation across the life-span may vary for the working and non-working impoverished whose eligibility for and access to health, educational and social resources differ. The experience of poverty may also differ for racial/ethnic groups such as Latinos/Hispanics, who despite overall poor social indicators, suffer few tobacco-related cancer disparities.
Education only partially explains current smoking and quitting among blacks/African Americans and American Indians/Alaska Natives compared to whites . High school dropout rates and misclassification of smoking are not likely to explain lower smoking rates among black/African American adolescents ; this may also be true for Latino/Hispanic youth who have the highest school dropout rates . Even after controlling for parental socio-economic status, black/African American youth are still less likely to smoke compared to other racial/ethnic groups [1,49]. However, the experience of education may differ. In 2002, blacks/African Americans earned the lowest average incomes compared to other racial/ethnic groups with equivalent educational attainment, and the income gap increased with increasing educational attainment . For blacks/African Americans, there may be an interactive effect of education–income–race/ethnicity across the tobacco disease continuum that may not be observed among some other racial/ethnic groups. In addition, because of the long history of oppression and racism in the United States against blacks/African Americans, one cannot rule out the differential effects of discrimination on blacks/African Americans compared to other minority racial/ethnic groups across the life-span.
Insufficient data are available to compare ETS exposure or hair cotinine levels among racial/ethnic groups, but some studies suggest that serum cotinine contributes to disparities through biological processes. Studies suggest that ETS exposure in children of light smokers is associated with the biologically effective dose of two known carcinogen–protein adducts (4-ABP and PAH) and general measures of genetic damage (SCEs), thereby suggesting that ETS exposure may cause molecular or genetic damage at any point in the life cycle . Blacks/African Americans may metabolize nicotine more slowly than whites and Latinos/Hispanics [55,58], and a number of gene polymorphisms are being examined (i.e. CPY2A6) which might explain differences in enzyme levels and activity and thus differences in cotinine levels [55,163–168]. However, small differences in the prevalence of CYP2A6 polymorphisms may not explain fully the differences between cotinine levels and health outcomes . Hypotheses about biological processes and gene polymorphisms may be critical steps to exploring alternative disease models for minority groups, but these hypotheses should be approached with caution; the quest for racial classification of genes will do little to reduce disparities among racial/ethnic groups.
Several caveats must be noted. A more comprehensive review of data on smoking among racial/ethnic groups prior to 1998 is reported in the 1998 Surgeon General Report. Psychosocial and cultural factors, community level influences, dependence/addiction, biological processes , menthol cigarettes, familial genes and susceptibility [170,171], gene polymorphisms and other factors may partially explain cancer disparities that exist, although we do not provide an exhaustive discussion of these factors. Such indicators are not normally part of surveillance systems and our goal here was to approach disparities from an epidemiological perspective using key national indicators. Several of these topics are discussed in other papers within this journal supplement. Biopsychosocial factors are discussed in Hammons et al. , dependence/addiction is discussed in Moolchan et al.  and community-based partnerships are discussed in Báezconde-Garbanati et al. . The cancer risks of smoking mentholated compared to non-mentholated cigarettes have not been demonstrated , but studies are examining how menthol contributes to carcinogenesis through increased toxicity, addictiveness and altered patterns of inhalation . Unemployment was not examined as a social variable; however, national estimates indicate that 35% of the unemployed smoke  and unemployment is higher among blacks/African Americans, American Indians/Alaska Natives and Latinos/Hispanics compared to whites. Unemployment has been associated with smoking-related mortality [178,179]. Tobacco marketing, type of cigarette smoked (i.e. light), other forms of tobacco, other drugs, alcohol or dietary factors may interact with cigarette use or occur concurrently. Use of other tobacco is low in most racial/ethnic groups . Among all cigarette users, 4.8% use smokeless tobacco . Geographic region was not a focus, but tobacco use in cities and neighborhoods may far exceed national averages. For example, 44% of those sampled in Harlem were current smokers . One final caveat is that we do not report specifically on LGBT racial/ethnic minorities because the epidemiological data are almost non-existent.
Approaches that address simultaneously TRHD, work-force diversity and empowerment, inequities in research and inequalities in access to care may provide the paradigm shift needed to accelerate progress in reducing disparities among minority racial and ethnic groups. Available data on indicators in the tobacco disease continuum challenge our assumptions about the dose–response model for minority racial/ethnic and low socio-economic status groups and it is clear that alternative models are needed to understand more clearly the relationship between tobacco use and exposure and health consequences among these groups. A number of recommendations have been made in previous reports on how to reduce tobacco-related health disparities among minority racial/ethnic groups [1,181], yet few recommendations have been implemented. It is not our intention to reiterate those recommendations, but to encourage collective action to adopt a framework which focuses simultaneously on disparities, diversity, inequities and inequalities. We therefore recommend the following:
1Support and create multi-ethnic and inclusive coalitions of researchers, practitioners, community organizations, families and policy makers who work collectively at the local, state and national levels to develop comprehensive solutions to eliminate TRHD in minority racial/ethnic groups.
2Develop long-term concrete plans on how to increase, support and retain the future pipeline of researchers, practitioners and advocates and support intergenerational collaborations and shared leadership among stakeholders.
3Work with policy makers, health systems, community service organizations, businesses and others to develop strategies that increase access, demand, availability and quality of care for underserved minority racial/ethnic groups.
4Develop research–community partnerships that help to strengthen our ability to address inequities in research, application, translation and dissemination.
5Work with coalitions at multiple levels to develop specific strategies to address social factors such as:
(a)reducing individual and community level poverty among minority racial/ethnic groups with the highest rates,
(b)increasing high school graduation rates among minority racial/ethnic groups and creating opportunities to support higher education, and
(c)developing strategies in work-places to protect workers in all social strata from environmental hazards.
While these recommendations merely scratch the surface, they serve to encourage collective action to eliminate the leading cause of preventable death in the United States and its causal factors. Furthermore, while research and public health efforts aim to reduce tobacco use among everyone, these recommendations serve as a specific call to action to eliminate disparities among minority racial and ethnic groups.
We extend our gratitude to the National Cancer Institute and the American Legacy Foundation who fund the Tobacco Research Network on Disparities. We thank Julie Volkman and D. B. Consulting for assisting with the preparation of the manuscript and Drs Cathy Backinger and Bradford Hesse for reviewing the manuscript.