Racial/ethnic differences in menthol cigarette smoking, population quit ratios and utilization of evidence-based tobacco cessation treatments
Steven S. Fu, Center for Chronic Disease Outcomes Research (CCDOR), VA Medical Center (152/2E), One Veterans Drive, Minneapolis, MN 55417, USA. E-mail: email@example.com
Aims This study examines the relationship between menthol cigarette smoking and the population quit ratio and whether menthol smokers differ in utilization of evidence-based smoking cessation aids among a nationally representative sample of US adult smokers.
Design, setting and particiants Secondary data analysis of cross-sectional data from the 2005 National Health Interview Survey (NHIS) Cancer Control Supplement. The NHIS is a nationally representative survey of US households conducted annually.
Measurements The main outcome variables of interest were (1) the population quit ratio and (2) use of smoking quit aids. All analyses were conducted using SAS version 9.2 with SUDAAN, which corrects for the complex sampling design of the study. Univariate analyses were used to determine variables that differed significantly by menthol status and utilization of types of quit aids. Multiple logistic regression analysis modeled the relationship between menthol smoking status, demographic characteristics and smoking-related characteristics on the population quit ratio and utilization of quit aids.
Findings We observed significant differences in the population quit ratio for menthol versus non-menthol among African American smokers (34% versus 49%, P < 0.001), but not among whites (52% versus 50%). In multiple logistic regression analysis, there was a significant interaction between race and menthol smoking status. African American menthol smokers were significantly less likely than white non-menthol smokers to have quit smoking (adjusted odds ratio: 0.72, 95% confidence interval: 0.53, 0.97) after controlling for age group, sex, marital status, region and average number of cigarettes smoked per day. Menthol smoking status was not associated with differences in utilization of quit aids.
Conclusions African Americans have the highest rates of menthol cigarette smoking of all racial and ethnic groups in the United States. Menthol cigarette smoking is associated negatively with successful smoking cessation among African Americans.
Tobacco use is the leading cause of death in the United States, responsible for more than 400 000 deaths annually . Certain demographic groups are more likely to suffer from the consequences associated with tobacco use, especially African Americans. Previous research has demonstrated that African American smokers are less likely to quit smoking, which is independent of socio-demographic factors and smoking-related characteristics [2–7]. The high rate of menthol cigarette smoking among African American smokers has been suggested as a possible contributor to the observed differences in tobacco cessation and existing tobacco-related health disparities between African Americans and whites . In the United States, 69% of African American smokers smoke menthol cigarettes compared with 23% of white smokers, 29% of Hispanic American smokers and 29% of Asian American smokers .
Menthol cigarettes currently comprise 20% of the tobacco industry's sales market, and advertising of these types of cigarettes are targeted specifically at women and racial/ethnic minorities . Menthol, a naturally occurring compound found in the peppermint plant, is used in cigarettes to give a mint or coolness sensation when smoking. The sales-weighted averages of menthol cigarettes are generally higher in tar and nicotine, and hence menthol cigarettes may be more addictive and more difficult to quit smoking compared with plain cigarettes . Research studies to date, however, are mixed on whether menthol cigarette smoking affects smoking cessation efforts such as quit attempts, length of quit attempts and smoking abstinence [11–18].
Effective tobacco dependence treatments are available, but racial/ethnic minority smokers are less likely than white smokers to use treatment, including pharmacotherapy and intensive behavioral counseling [7,19,20]. It is unclear from previous research if menthol smokers are less likely to utilize quit aids to stop smoking. This study examined the relationship between menthol smoking and the population quit ratio and whether menthol smokers differed in utilization of evidence-based smoking cessation aids, controlling for several demographic and smoking characteristics among a nationally representative sample of US adult smokers.
Data source and population
This study analyzed data from the 2005 National Health Interview Survey (NHIS). This nationally representative study is conducted annually by the National Center for Health Statistics, Centers for Disease Control and Prevention. Details of the NHIS have been published elsewhere [21,22]. Briefly, the NHIS is a cross-sectional survey of nationally representative non-institutionalized adults. The NHIS uses a multi-stage sampling plan to collect a variety of health information via a computer-assisted personal interview. In 2005, a Cancer Control Supplement was used that collected information related to a variety of cancer-related health indicators, including smoking and cessation utilization on a subsample of the adult population.
This study used the sample adult file consisting of 31 428 people aged 18 and older. Approximately 42% of adults reported being either a current (n = 6511) or former (n = 6774) smoker (Fig. 1). To determine menthol cigarette use, both current and former smokers were asked if their usual cigarette brand was mentholated. Individuals missing information related to menthol cigarette status were not included in further analyses. The sample size was limited to 12 004 adults for which menthol cigarette status was known. Approximately 26% of current smokers used menthol cigarettes and 22% of former smokers used menthol. In the NHIS Cancer Control Supplement current smokers were asked if they had tried to quit smoking in the past 12 months and whether they had tried a particular quit aid (i.e. nicotine gum, patch, spray, etc.). All current smokers with known menthol status were included in both sets of analyses. All former smokers with known menthol status were included in the analyses of the population quit ratio. For consistency, in the analyses of the utilization of quit aids, former smokers were limited to individuals who had reported quitting smoking within the previous 12 months. About 32% of current smokers who tried to quit in the past year reported using a quit aid and 25% of past-year former smokers reported using a quit aid.
The main outcome variables of interest for the current analysis were (1) the population quit ratio and (2) utilization of quit aids. The population quit ratio was calculated by dividing the total number of former smokers by the total number of individuals who had reported smoking during their life-time (i.e. both former and current smokers). Subanalyses were conducted by racial/ethnic categories and menthol smoking status. To determine whether a smoker utilized quit aids for smoking cessation, former smokers were asked the following question: ‘Thinking back to when you stopped smoking completely, did you use ANY of the following: . . . ?’. Answer choices included nicotine gum, patch or spray, inhaler or lozenge; prescriptions such as Zyban, buproprion or Wellbutrin; telephone quit line, a support group or the internet. Current smokers were asked: ‘Thinking back to when you tried to QUIT smoking in the PAST 12 MONTHS, did you use any of the following: . . . ?’, and answer choices were the same as for former smokers. Responses were coded dichotomously (0 = did not use, 1 = used).
Predictor variable and covariates
Menthol cigarette status was assessed by self-report of whether or not respondent's usual brand of cigarettes was mentholated (0 = non-menthol, 1 = menthol). The specific question asked: ‘Is (was) your usual cigarette brand menthol or non-menthol?’. Additional smoking characteristics included in this study were the average age when the respondent first began smoking regularly and the average number of cigarettes smoked per day (cpd) (former smokers were asked to think back to when they smoked regularly and report how many cigarettes they usually smoked each day); both measures have been found to be associated with nicotine dependence [23,24]. Current smokers were also asked whether they had attempted to quit smoking within the past year. We examined outcomes by several demographic characteristics, including age, sex, race [white, African American, American Indian/Alaskan Native (AI/AN) and Asian], Hispanic ethnicity, marital status, educational status (high school graduate or less versus some college or more) and region of the United States (South, Northeast, Midwest and West).
All analyses were run using SAS version 9.2 with SUDAAN to account for the complex weighting structure of the survey. Pearson's χ2 tests of independence and Student's t-tests were conducted to determine significant demographic and smoking differences between menthol and non-menthol cigarette users and use of quit aids for both current and former smokers. Results were considered significant with a P-value of less than 0.05 adjusted with a Bonferroni correction for multiple comparisons. Multiple logistic regression analyses were run to model the association between menthol cigarette use and the population quit ratio and the utilization of quit aids controlling for significant demographic and smoking characteristic variables.
The demographic and smoking characteristics of current and former menthol and non-menthol smokers are presented in Table 1. Overall menthol smoking prevalence was significantly different by sex, region of the United States, race, marital status and average number of cigarettes smoked per day for both current and former smokers and age for former smokers only. African American smokers reported the highest prevalence of menthol smoking compared with other racial groups for both current and former smokers (76% and 63%, respectively). Neither Hispanic ethnicity, education nor the average age when a person initially began smoking regularly appeared different for menthol versus non-menthol smokers.
Table 1. Demographic and smoking characteristics of current and former smokers by menthol status.
|Sex|| || || || || || || || |
| Female||2987||33.1||67.0|| ||2891||33.3||66.7|| |
|Region|| || || || || || || || |
| Northeast||984||33.9||66.1|| ||1103||24.9||75.1|| |
| South||2291||28.9||71.1|| ||2044||27.7||72.3|| |
| West||1142||16.3||83.7|| ||1317||19.9||80.1|| |
|Hispanic ethnicity|| || || || || || || || |
| No||5213||27.3||72.7|| ||5300||25.0||75.0|| |
|Race|| || || || || || || || |
| White||4932||20.2||79.8|| ||5147||21.3||78.7|| |
| African American||861||76.0||24.1||<0.001*||573||63.4||36.6||<0.001*|
| AI/ANa||54||16.4||83.6|| ||45||27.6||72.4|| |
| Asian||119||32.9||67.1|| ||98||29.8||70.2|| |
|Marital status|| || || || || || || || |
| Not married||3752||29.1||70.9|| ||2770||28.5||71.5|| |
|Education|| || || || || || || || |
| High school or less||3425||28.3||71.7|| ||2797||25.0||75.0|| |
| Some college/degree||2589||25.4||74.6||0.025||3114||24.5||75.6||0.677|
|Age group (years)|| || || || || || || || |
| 18–24||707||31.6||68.4|| ||203||33.8||66.2|| |
| 45–64||2212||27.9||72.1|| ||2374||28.3||71.7|| |
| 65 or more||521||22.6||77.4|| ||1960||19.3||80.7|| |
|Average no. of cpd||16.8||14.6||17.5||<0.001*||18.6||16.8||19.1||<0.001*|
|Average age first smoked regularly (years)||19.7||19.6||19.7||0.714||21.1||21.4||21.0||0.495|
For current and former smokers, non-menthol smokers reported a higher number of cigarettes smoked per day on average than menthol smokers (current 17.5 versus 14.6; former 19.1 versus 16.8, respectively). The average age when both current and former smokers first began to smoke regularly did not differ significantly between menthol and non-menthol smokers.
In addition, approximately 43% of current smokers attempted to quit in the past year and of current menthol smokers, 49% reported a quit attempt in the past year, while 41% of non-menthol smokers reported a quit attempt (data not shown). Additionally, past-year quit attempts did not appear to differ by Hispanic ethnicity, but did by racial category (P = 0.008), with 49% of African American smokers and 48% of Asian American smokers reporting a quit attempt in the previous year versus 41% of white smokers and AI/AN smokers.
The quit ratio for white smokers was significantly higher than the quit ratio for African American smokers (51% versus 38%, P < 0.001), but not for Asian American smokers (51% versus 41%) or AI/AN smokers (51% versus 39%) (data not shown). In addition, the quit ratios were significantly higher for non-Hispanics versus Hispanics (50% versus 44%, P < 0.001) and non-menthol versus menthol smokers (50% versus 47%, P = 0.014). When examining the quit ratio by menthol status and race or ethnicity, there is no significant difference in the quit ratios for menthol versus non-menthol smokers for whites (52% versus 50%), Asian Americans (38% versus 42%), AI/AN (52% versus 35%) or Hispanics (40% versus 45%). However, we observed significant differences in the quit ratio for menthol versus non-menthol among African American smokers (34% versus 49%, P < 0.001). In multiple logistic regression analysis, there was a significant interaction between race and menthol smoking status (Table 2). African American menthol smokers were significantly less likely than white non-menthol smokers to have quit smoking (adjusted odds ratio: 0.72, 95% confidence interval: 0.53, 0.97, P-value 0.031) after controlling for age group, sex, region, marital status and average number of cigarettes smoked per day.
Table 2. Predictors of population quit ratio.
|Sex|| || || |
| Male (ref)||1.00|| || |
| Female||1.03||(0.93, 1.14)||0.529|
|Region|| || || |
| Northeast (ref)||1.00|| || |
| Midwest||0.81||(0.70, 0.95)||0.009|
| South||0.80||(0.68, 0.94)||0.007|
| West||1.17||(1.00, 1.38)||0.047|
|Marital status|| || || |
| Married (ref)||1.00|| || |
| Not married||0.50||(0.45, 0.56)||<0.001|
|Age group (years)|| || || |
| 18–24||0.09||(0.07, 0.11)||<0.001|
| 25–44||0.13||(0.11, 0.15)||<0.001|
| 45–64||0.24||(0.21, 0.27)||<0.001|
| 65 or more||1.00|| || |
|Average number of cpd||1.00||(1.00, 1.00)||0.814|
|Interaction of menthol status and race|
| Non-menthol (ref) × white (ref)||1.00|| || |
| Menthol × African American||0.72||(0.53, 0.97)||0.031|
| Menthol × AI/AN||2.00||(0.56, 7.23)||0.287|
| Menthol × Asian||0.74||(0.37, 1.48)||0.396|
Utilization of quit aids differed significantly for current smokers based upon several demographic and smoking characteristics, including sex, race and Hispanic ethnicity, marital status, education, current age, average number of cigarettes per day and average age first began smoking regularly for current smokers (Table 3). For past-year former smokers, only Hispanic ethnicity, current age and average number of cigarettes per day explained differences between whether or not a person utilized quit aids (Table 3). AI/AN current smokers report the lowest utilization of quit aids than other racial groups (17.4%, P < 0.001). Current and past-year former smokers who used quit aids appeared to smoke more cigarettes per day on average than smokers not reporting use of quit aids (current smokers: 17.2 versus 13.7 cpd; former smokers: 22.1 versus 12.9 cpd).
Table 3. Demographic and smoking characteristics of participants that use quit aids by smoking status.
|Sex|| || || || || || |
| Female||35.9||64.1|| ||30.1||69.9|| |
|Region|| || || || || || |
| Northeast||38.8||61.2|| ||27.1||72.9|| |
| South||31.1||68.9|| ||25.1||74.9|| |
| West||32.5||67.5|| ||22.4||77.6|| |
|Hispanic ethnicity|| || || || || || |
| No||33.5||66.5|| ||28.6||71.5|| |
|Race|| || || || || || |
| White||34.3||65.7|| ||27.7||72.3|| |
| African American||21.6||78.4||<0.001*||15.5||84.5||0.115|
| AI/ANa||17.4||82.7|| ||20.0||80.0|| |
| Asian||30.0||70.0|| ||12.5||87.6|| |
|Marital status|| || || || || || |
| Not married||28.0||72.0|| ||23.9||76.1|| |
|Education|| || || || || || |
| High school or less||28.5||71.5||<0.001*||24.1||75.9||0.370|
| College||36.6||63.4|| ||27.9||72.1|| |
|Age group (years)|| || || || || || |
| 18–24||21.3||78.7|| ||5.1||95.0|| |
| 45–64||39.1||60.9|| ||37.0||63.0|| |
| 65 or more||30.1||69.9|| ||28.5||71.5|| |
|Average no. of cpd||17.2||13.7||<0.001*||22.1||12.9||<0.001*|
|Average age first smoked regularly||18.0||19.7||<0.001*||19.2||22.3||0.050|
Approximately 27% of current smokers who tried to quit in the past year and 22% of past-year former smokers used some form of nicotine replacement therapy to help them stop smoking (data not shown). The nicotine patch was the most commonly used form of nicotine replacement therapy (NRT), with 19% of current smokers trying the patch and 16% of past-year former smokers successfully quitting smoking using it. Overall, 32% of current smokers who tried to quit in the past 12 months tried some form of quit aid to help them stop smoking and 26% of past-year former smokers successfully used some form of quit aid to help them stop smoking. Approximately 93% of all current smokers reported that a doctor had advised them to quit smoking cigarettes or other kinds of tobacco products within the past 12 months, while 16% reported a dentist had advised them to quit (data not shown). These proportions did not differ by whether the person reported smoking menthol or non-menthol cigarettes.
The overall odds of using any type of quit aid did not differ significantly between menthol or non-menthol smokers for either current or past-year former smokers after controlling for demographic and smoking characteristics (Table 4). These results were similar when specific groups of quit aids were examined separately, e.g. NRT only or NRT combined with prescription drugs. Because menthol cigarette status was not associated with utilization of quit aids, no interaction term between race and menthol smoking was included in this model. Females reported significantly higher odds than males of utilizing quit aids for both current and former smokers (P < 0.001 and P = 0.008, respectively). African American current smokers reported significantly lower odds of utilizing quit aids than white current smokers (P = 0.025). In addition, living in the Midwest or South or not being married significantly decreased the odds of utilizing quit aids among current smokers but not former smokers. College education increased the odds of utilizing quit aids by 60% among current smokers (P < 0.001). On average, for every one cigarette increase in the average number of cigarettes smoked per day, the odds of utilizing quit aids increased by 5% for current smokers (P < 0.001) and 6% for former smokers (P < 0.001).
Table 4. Predictors of utilization of quit aids for current and former smokers.
|Use of any quit aid|| || || || || || |
| Menthol status|| || || || || || |
| Non-menthol (ref)||1.00|| || ||1.00|| || |
| Menthol||1.05||(0.80, 1.36)||0.734||1.29||(0.74, 2.26)||0.374|
| Sex|| || || || || || |
| Male (ref)||1.00|| || ||1.00|| || |
| Female||1.60||(1.25, 2.04)||<0.001||1.88||(1.18, 2.99)||0.008|
| Region|| || || || || || |
| Northeast (ref)||1.00|| || ||1.00|| || |
| Midwest||0.57||(0.38, 0.84)||0.005||1.16||(0.59, 2.31)||0.662|
| South||0.66||(0.46, 0.95)||0.024||1.08||(0.52, 2.25)||0.830|
| West||0.79||(0.50, 1.25)||0.311||1.07||(0.49, 2.31)||0.869|
| Race|| || || || || || |
| White (ref)||1.00|| || ||1.00|| || |
| African American||0.64||(0.43, 0.94)||0.025||0.54||(0.21, 1.34)||0.180|
| AI/AN||0.55||(0.14, 2.22)||0.404||2.10||(0.25, 17.65)||0.492|
| Asian||0.98||(0.32, 3.03)||0.977||0.53||(0.12, 2.31)||0.398|
| Hispanic ethnicity|| || || || || || |
| Yes (ref)||1.00|| || ||1.00|| || |
| No||1.47||(0.94, 2.32)||0.094||3.89||(1.60, 9.41)||0.003|
| Marital status|| || || || || || |
| Married (ref)||1.00|| || ||1.00|| || |
| Not married||0.70||(0.55, 0.89)||0.004||1.17||(0.73, 1.85)||0.514|
| Education|| || || || || || |
| High school or less (ref)||1.00|| || ||1.00|| || |
| College||1.60||(1.25, 2.04)||<0.001||1.39||(0.86, 2.24)||0.174|
| Age group (years)|| || || || || || |
| 18–24||0.71||(0.39, 1.30)||0.267||0.15||(0.05, 0.49)||0.002|
| 25–44||0.99||(0.61, 1.61)||0.965||1.38||(0.64, 2.94)||0.407|
| 45–64||1.14||(0.71, 1.86)||0.585||1.52||(0.68, 3.42)||0.309|
| 65 or more (ref)||1.00|| || ||1.00|| || |
| Average number of cpd||1.05||(1.03, 1.07)||<0.001||1.06||(1.04, 1.09)||<0.001|
| Average age first started smoking||0.98||(0.96, 1.01)||0.164||0.99||(0.97, 1.01)||0.258|
In this analysis of the 2005 National Health Interview Survey, we found that race significantly modifies the effect of menthol cigarette smoking on smoking cessation as measured by the population quit ratio. Among African Americans, menthol cigarette smoking is associated with decreased likelihood of smoking cessation. Among Hispanics, the quit ratio was lower for menthol smokers compared with non-menthol smokers, but this difference was not statistically significant. Menthol cigarette smoking was not observed to significantly decrease the likelihood of smoking cessation among whites, Asians or AI/AN. Menthol smoking status was also not found to be associated with differences in utilization of evidence-based tobacco cessation aids. Several demographic characteristics were associated with a lower likelihood of using quit aids, including African American race, living in certain regions of the United States and not being married.
Our findings provide further support of emerging research that is suggestive of a negative impact of menthol smoking on smoking cessation among racial and ethnic minorities, particularly African Americans. Okuyemi and colleagues' study  is the first of four studies to our knowledge to report a negative impact of menthol smoking on smoking abstinence in African Americans. In this secondary analysis of a randomized placebo-controlled trial of bupropion sustained-release conducted specifically among African Americans, menthol smoking was found to attenuate the effect of bupropion on cessation among African American smokers less than 50 years of age. Secondly, Okuyemi and colleagues  also analyzed data from a clinical trial assessing the efficacy of nicotine gum and motivational interviewing counseling among African American light smokers. Biochemically verified 7-day point prevalence smoking abstinence rates were lower among African American menthol smokers compared with African American non-menthol smokers (11.2% versus 18.8%, P-value = 0.015). Thirdly, in an observational cohort analysis of a diverse sample of smokers (n = 1688) at a single specialized tobacco cessation treatment clinic, within racial/ethnic group, African Americans and Latino menthol smokers were less likely to be abstinent at 6-month follow-up compared with their non-menthol smoking counterparts . Among white smokers, menthol smoking status was not associated with smoking abstinence. Fourthly, a cross-sectional analysis of the 2005 National Health Interview Survey, using a restricted sample compared with the current analysis, also reported finding an interaction between race/ethnicity and menthol smoking status . Among smokers who ever attempted to quit and did not use other tobacco products, menthol smoking was associated with decreased likelihood of being a former smoker for non-whites (African Americans and Hispanics); however, for whites, menthol smoking was associated with increased likelihood of being a former smoker. In contrast, three other studies (two community-based studies and one clinical study) have found no association between use of menthol cigarettes and cessation [16–18]. It is possible that demographic differences in the populations studied may explain the differences in the outcomes across studies. For example, the two null studies by Muscat and colleagues  (80% >45 years of age) and Fu and colleagues  (77% >50 years) included predominantly older adults, whereas studies showing negative effects of menthol included younger African American smokers.
Several explanations have been proposed as potential factors for why menthol cigarette smoking may disrupt cessation attempts. For example, menthol smokers may achieve higher levels of nicotine and nicotine metabolites (e.g. cotinine) which may be due to greater inhalation per puff because of the soothing and anesthetic effects of menthol . Furthermore, menthol cigarette smoking has been shown to inhibit nicotine metabolism resulting from slower oxidative metabolism of nicotine to cotinine and by slower glucuronide conjugation . Inhibition of nicotine metabolism by menthol cigarette smoking may contribute to higher nicotine dependence and hence more difficulty with quitting smoking. In addition, the metabolism of cotinine is slower among African Americans , which might provide a basis for the observation that the effects of menthol smoking on abstinence is most striking among African Americans compared with other ethnic groups. Alternatively, the observed effects of menthol smoking on smoking abstinence may be due to self-selection bias of the types of African American smokers who choose to smoke menthol cigarettes. However, in the current analysis, we adjusted for several demographic factors including age group, sex, region of the United States and education and the observed effects persisted.
We observed that menthol smokers were significantly more likely to have a past-year quit attempt than non-menthol smokers, which indicates that menthol smokers, compared with non-menthol smokers, are not less motivated to quit smoking and thus engage in quit smoking attempts. However, menthol smokers were less likely to quit smoking (based on the population quit ratio). In addition, despite being more likely to make a quit attempt in the past year, African Americans were less likely to succeed. Our analysis also found that consistent with previous research, African Americans are less likely to utilize quit aids during attempts to quit smoking [7,20,28]. The lower utilization of quit aids by African Americans may be a contributing factor to observed lower cessation rates of African Americans .
Several strengths mark this study, including the large, national representative sample size and detailed questions regarding menthol cigarette use. This study also examined all ever smokers, not just those smokers who have attempted to quit, which is relevant because menthol cigarette smoking may inhibit ever engaging in a quit attempt. This study, however, has several limitations, including self-reported measures for both menthol cigarette use and use of quit aids, which may result in misclassification. Findings for Hispanics, Asian Americans and AI/AN may also have limited generalizeability, resulting from selection bias due to challenges in including these racial and ethnic groups in large national surveys. In addition, as the study is cross-sectional, we can only assess associations between menthol cigarette use and cessation or use of quit aids.
In conclusion, African Americans have the highest rates of menthol cigarette smoking compared with other racial and ethnic groups. Among African Americans, menthol cigarette smoking is associated significantly with decreased likelihood of smoking cessation as measured by the population quit ratio. This study, combined with previous literature, provides further evidence and support for the finding that menthol cigarette smoking negatively affects the achievement of successful smoking cessation among racial/ethnic minorities, especially younger African Americans. To date, no studies in the published literature have been specifically designed to assess prospectively the effect of menthol cigarette smoking on smoking cessation. Prospective studies and further research are needed to answer definitively whether menthol cigarette smoking affects smoking cessation negatively among African Americans and other racial/ethnic minorities.
Declarations of interest
This material is based on work supported in part by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development and Health Services Research and Development. The views expressed in this paper are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government.