Cigarette Prices and Disparities in Smoking Cessation in the United States

Abstract

Introduction

Achieving cessation in people with established smoking patterns remains a challenge. Increasing cigarette prices has been one of the most successful strategies for lowering smoking rates. The extent to which it has remained effective in encouraging cessation among adults in recent years and how the effectiveness has varied by sociodemographic characteristics is unclear.

Aims and Methods

Using repeated cross-sectional data collected by the Tobacco Use Supplement of the Current Population Survey, we investigate the relationship between cigarette prices and cessation from 2003 to 2019 in adults at least 25 years old. We examine the associations between price and cessation in the population overall and by sex, race and ethnicity, and socioeconomic status.

Results

We found mixed support for associations between greater local prices and cessation. Unadjusted models showed that greater local prices were associated with greater odds of cessation, but the associations did not persist after controlling for sociodemographic characteristics. The associations did not significantly differ by respondent characteristics. Sensitivity analysis using alternative specifications and retail state price as the main predictor showed similar results. Sensitivity analysis with controls for e-cigarette use in the 2014–2019 period showed that greater local price was associated with cessation among adults with less than a high school degree. When stratified by year of data collection, results show that greater local prices were associated with cessation after 2009.

Conclusions

Overall, the study adds to the conflicting evidence on the effectiveness of increasing prices on smoking cessation among adults with established smoking patterns.

Implications

Higher cigarette prices have been one of the most successful tools for lowering smoking prevalence. It remains unclear how effective they have been in recent years in encouraging adults with established smoking patterns to quit. The study’s results show that greater local prices were associated with higher odds of cessation, but the association did not persist after sociodemographic adjustment. In a sensitivity analysis, greater local price was associated with cessation among people with less than a high school degree in models controlling for e-cigarette use. We also found evidence that greater local price was associated with cessation after 2009. More comprehensive smoke-free coverage was also associated with greater odds of cessation. The study’s results highlight that encouraging cessation among adults with an established smoking pattern remains a challenging policy problem even when cigarette prices rise.

Introduction

Tobacco control efforts at the federal, state, and local levels have translated to the lowest smoking rate in the United States since the mid-1960s.¹ In 2020, only 13% of US adults smoked.² Research shows that raising tobacco prices via higher taxes has played a central role in tobacco control’s success, estimating that each 10% increase in price has translated to about a 4% decrease in cigarette demand in the United States.³ In 2023, the tax burden on cigarettes in the United States continues to lag behind the World Health Organization recommendation of 75% of the product retail price,^4,5 and it varies greatly among US states and local areas.⁶ For example, in New York City, a person who smokes pays a $1.50 local tax in addition to the $5.35 state tax and $1.01 federal tax on each standard pack of cigarettes.⁷ The same person is only subject to a $1.34 state tax and $1.01 federal tax in Florida.⁸

Smoking rates have remained high among Americans with the fewest socioeconomic resources, among some racial and ethnic minoritized populations, and among people living with other disadvantages, such as mental health conditions.^9–11 An adult with a GED is five times more likely to smoke than one with a college degree (35% vs. 7%).⁹ Similarly, an adult living in a household with an annual income of $35 000 or less is three times more likely to smoke than one living in a household with $100 000 or more (21% vs. 7%).⁹ People who identify as American Indian or Alaska Native are more likely to smoke cigarettes than non-Hispanic white adults (21% vs. 16%).⁹

Raising cigarette prices is thought to help lower smoking rates through two mechanisms: first, higher prices discourage smoking initiation¹² among youth who do not regularly smoke. Young adults and teenagers tend to be more sensitive to price changes compared to older age groups.^13,14 They typically have lower income levels and are more influenced by fluctuations in prices. As the cost of cigarettes rises, it becomes less enticing for young adults to purchase them, leading to a decrease in consumption. Lesser availability of cigarettes among young people will translate to fewer opportunities to experiment and a lower likelihood of transition from experimentation to established smoking.¹⁵ Second, they promote cessation among people with established smoking patterns by making it more expensive to maintain addiction. The literature focusing on the second mechanism is the less voluminous of the two, only infrequently focusing on adults who smoke long-term.¹⁴ Using the 1998 and 1999 waves of the Tobacco Use Supplement to the Current Population Survey (TUS–CPS), Levy and colleagues found that higher prices were associated with an increased probability of making a quit attempt and staying abstinent for at least 3 months.¹⁶ The likelihood of success varied by the characteristics of the participants. Adults who smoked and were older or had higher socioeconomic status (SES) were more likely to succeed in their quit attempts. In contrast, Messer and colleagues,¹⁷ who examined the TUS–CPS repeated cross-sectional data collected between 1992 and 2002 and focused on the effects of policy changes in California, found no increased probability of successful cessation associated with higher prices among adults who smoked and were older than thirty-four. Messer further examined the age gradient in a follow-up study, demonstrating that cessation among adults younger than 30 is much more common than cessation later in life.¹³ In a unique natural experiment, Dunbar and colleagues took advantage of the compulsory relocation of US military service members to installations in 2011 and evaluated the associations between changes in cigarette taxes and individual smoking. Their results show a higher likelihood of cessation after re-location to areas with higher taxes.¹⁸ Overall, the results of studies on smoking prices, taxes, and cessation among people with an established smoking pattern show heterogeneity across samples, time periods, and analytic strategies.

Our study investigates the relationship between cigarette price and cessation in US adults, focusing on the present era of higher taxes and prices in some US areas. A systematic review of studies examining variation area-level prices has found that prices tend to be slightly lower in areas with lower median income and a higher share of youth and African Americans.¹⁹ Given the persistently high levels of smoking addiction in socially disadvantaged groups, the question of differential sensitivity to changes in prices by sociodemographic characteristics has become even more urgent. Prior evidence shows inconsistent patterns of the effects of tobacco control interventions on smoking cessation and smoking prevalence in low SES populations.^20,21 It also shows that self-reported cigarette purchase prices vary by race and ethnicity in the United States. Non-Hispanic white and American Indian individuals report paying the lowest cigarette prices. Black, Hispanic, and Asian individuals pay higher prices on average, at an estimated $0.42, $0.68, and $0.89 higher price per pack, respectively, more than white individuals after the adjustment for other sociodemographic and local area characteristics.²² The current study aims to fill the gap in our understanding of the relationship between cigarette prices and smoking cessation between 2003 and 2019. We focus on the potential heterogeneous treatment effects of increased prices on smoking cessation by sex, race and ethnicity, and SES. Using the TUS–CPS,²³ we ask: (1) Have higher levels of local cigarette prices been associated with greater probabilities of cessation? (2) Has the effectiveness of higher cigarette prices in encouraging cessation varied by sociodemographic characteristics?

Methods

Population

We used data from five cross-sectional waves of the TUS–CPS (2003, 2006–2007, 2010–2011, 2014–2015, and 2018–2019), which we downloaded from the IPUMS data repository.²⁴ The CPS collects data about employment and labor force participation from American residents in noninstitutionalized households by a rotating monthly panel design. The TUS supplement covers the use of tobacco products, personal tobacco-use history, and attitudes toward tobacco use and tobacco control policies.²³ The supplementary data are gathered three times from three different groups of respondents within each supplement wave. We restricted our analytic sample to self-respondents at least 25 years old (n = 753 021) who reported they smoked a year ago (n = 134 441). Respondents who reported they started smoking within the last 2 years and who did not report whether they smoked a year ago (n = 124) were excluded. Because local areas where smoking was rare were likely systematically different, and the typical price cannot be accurately estimated from small subsamples, we did not use local areas with fewer than 20 sampled smokers (n = 32 605). The final analytic sample includes 101 712 respondents.

Measures

Smoking Cessation Outcome Variable

The cessation outcome was operationalized as a recent successful smoking cessation. We classified respondents as having achieved successful cessation if they had smoked 12 months before the interview and reported that they quit at least 90 days before the day of the survey.²⁵

Cigarette Price

Similar to recent work by Kalousova et al.²⁶ and Pesko et al.,²⁷ we calculated local cigarette prices at the county level or the core-based statistical area (CBSA), a statistical geographic entity consisting of the county or counties associated with at least one urbanized area/urban cluster of at least 10 000 population, plus adjacent counties with a high degree of social and economic integration²⁸ for each month and year. We used individual reported prices for the last pack or carton purchased after applying any coupons or discounts. The individual contributions to the local mean were weighted by self-reported smoking intensity so that a smoker who purchased more cigarettes contributed more to the local price mean. If a local county mean price was available, an individual was assigned their county mean as their local county price. If a county marker was not available or there were fewer than twenty sample members who smoked in their county, CBSA values were used. When the CBSA marker was not available either, or there were fewer than twenty people who smoked in the CBSA, the average for the remainder of the state was used. Throughout the text, we refer to this calculated price as “local.” For 18.6% of the sample, local price refers to the average county price; for 40.0% of the sample, local price is aggregated at the CBSA level, and for 41.3%, local price was aggregated at the state level from observations not assigned to either a county or a CBSA. Observations reporting a price below $0.39, ie less than the federal tax (n = 131), before 2017 and $1.01 after 2017 (n = 541), were not used. Finally, we adjusted the price for inflation using the Consumer Price Index (CPI), so that all-prices are in constant 2019 dollars.²⁹

Sociodemographic Variables

The likelihood of successful cessation varies by sociodemographic characteristics.³⁰ The regression models control for sex (men or women), age measured in years (continuous, ≥25), race and ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, non-Hispanic other race and ethnicity), education (<high school, high school diploma or GED, some college, four-year college or graduate school), and annual household income (<$15k; $15k–29 999; $30k–49 999; $50–74 999; ≥$75k), in constant 2019 dollars.²⁹ All-individual sociodemographic characteristics were self-reported.

Smoke-Free Law Coverage

We measured the smoke-free law coverage of each state by year and month from the American Nonsmokers’ Rights Foundation database (ANRF).^25,31 A dichotomous variable indicated whether 100% of the state’s population was covered by smoke-free laws in workplaces and hospitality venues (restaurants and bars).

Statistical Analysis

We estimated logistic regression models predicting cessation by price, adjusting for age (linear and quadratic form), sex, race and ethnicity, education, annual household income, state-level smoke-free laws, and fixed effects for state and year. We tested effect modification by sex, race and ethnicity, education, and income on the multiplicative scale by including interaction terms in separate models. We evaluated models with three-way interactions between price, sex, and race and ethnicity; between price, sex, and education; and between price, sex, and income to test if the price measure’s effectiveness for each category of the sociodemographic factors varied by sex. We also plotted predicted marginal probabilities of smoking cessation for those interaction models. We applied the Benjamini–Hochberg correction method with the false discovery rate at 5% across all-interaction models to adjust for multiple testing.

About 5.1% of the final analytic sample had missing household income. We imputed the missing income values using chained equations with IVEware software version 0.3.³² The results presented are based on analysis that was performed on the five imputed datasets generated by the imputation procedure.

In sensitivity analyses, we additionally controlled for interactions between sociodemographic characteristics and indicators for state and year to account for state-specific disparity patterns. The interactions were not included in the main models because of small sample sizes for several race and ethnicity and state combinations. We also: (1) re-estimated all models using the average retail price per cigarette pack measured at the state level from the Tax Burden on Tobacco 2003–2018 data⁸ to assess whether the associations differ from those using our measure and (2) stratified into periods before 2009 and after 2009 to potentially capture the difference because of the Family Smoking Prevention and Tobacco Control Act. In years when e-cigarette use reports were available, TUS 2014–2015 and 2018–2019, we estimated models controlling for current use. We used the question “Have you ever used e-cigarettes even one time?” and a follow-up among people who ever used e-cigarettes that asked, “Do you now use an e-cigarette every day, some days, or not at all?’ If the respondent answered other than “not at all,” they were classified as having current e-cigarette use. All analyses were adjusted for replicate weights to account for the survey design and conducted in Stata SE, version 15.³³

Results

Table 1 shows the weighted descriptive statistics of the analytic sample. We measured that 6.7% of US adults 25 and older who smoked a year before their interview reported recent successful cessation in 2003, 7.4% in 2006–2007, 7.1% in 2010–2011, 8.4% in 2014–2015, and 8.3% in 2018–2019. These estimates align closely with data reported by the Centers for Disease Control and Prevention.³⁴ Over the study period, about 32% of respondents lived in a state with 100% smoke-free coverage in workplaces and hospitality venues, increasing from 3.9% in 2003 to 60.5% in 2018–2019. Age, sex, race and ethnicity, education, and CPI-adjusted family income were similar across the five study waves. For the reader’s convenience, we also present the descriptive characteristics stratified by whether or not a person achieved successful cessation in Supplementary Appendix Table 1.

Table 1.

Weighted Descriptive Characteristics of the CPS-TUS Analytic Sample

	Overall	2003	2006–2007	2010–2011	2014–2015	2018–2019
Age categories
% 25–39	35.7	36.7	36.5	35.7	34.9	33.5
% 40–54	36.9	40.5	39.4	37.3	33.0	29.8
% ≥55	27.4	22.8	24.1	27.0	32.1	36.6
% Men	54.1	53.7	54.0	54.1	53.9	55.5
Race and ethnicity
% Non-Hispanic white	72.6	75.2	74.5	73.1	69.2	68.1
% Non-Hispanic black	12.0	11.3	11.1	11.9	13.7	13.0
% Hispanic	9.7	8.5	9.4	9.4	10.7	11.6
% Other non-Hispanic	5.6	5.0	4.9	5.7	6.3	7.2
Education
% Less than high school	16.5	17.1	17.5	16.0	16.6	14.4%
% HS graduate	37.9	38.7	38.1	38.6	36.9	36.3
% Some college	30.2	28.9	29.0	30.3	31.7	32.7
% ≥College	15.3	15.2	15.4	15.0	14.8	16.5
CPI-adjusted income
% $0–14 999	21.1	24.2	20.5	20.5	21.6	17.1
% $15 000–29 999	24.6	33.2	24.0	22.9	20.6	17.2
% $30 000–49 999	24.4	24.0	28.4	24.1	22.1	21.1
% $50 000–74 999	14.3	7.5	13.2	17.3	18.2	19.5
% ≥$75 000	15.6	11.1	13.9	15.2	17.6	25.1
100% Smoke-free coverage	32.4	3.9	21.7	45.5	50.6	60.5
% Recent cessation	7.5	6.7	7.4	7.1	8.4	8.3
Price per pack local area in CPI-adjusted dollars	3.84 (0.01)	2.23 (0.01)	2.80 (0.01)	4.41 (0.01)	5.09 (0.01)	6.22 (0.02)
N	101 712	28 404	25 060	21 330	16 536	10 382

Table 2 presents bivariate and adjusted logistic regression models predicting recent successful smoking cessation, ie 90-day smoking cessation among respondents who were smoking 12 months ago. We observed a statistically significant positive association between local price and recent successful cessation in the bivariate model (OR: 1.08, 95% CI = 1.02 to 1.15). However, the association was not statistically significant in the fully adjusted model (OR: 1.04, 95% CI = 0.98 to 1.11), suggesting that differences in sociodemographic characteristics may explain the association between local cigarette price and recent cessation. Relative to non-Hispanic white adults, non-Hispanic black adults and non-Hispanic members of other racial and ethnic groups had lower odds of achieving recent successful cessation (OR: 0.78, 95% CI = .69 to 0.87; OR: 0.84, 95% CI = 0.71 to 0.98, respectively). Respondents with less than a high school diploma, a high school diploma, and some college education had lower odds of recent successful cessation compared to college graduates (OR: 0.52, 95% CI = 0.47 to 0.59; OR: 0.59, 95% CI = 0.54 to 0.64; OR: 0.80, 95% CI = 0.74 to 0.86, respectively). Individuals with annual family income <$15k, $15 000–29 999, and $30 000–49 999 had lower odds of recent successful cessation compared to the highest income group of $75 000 + (OR: 0.67, 95% CI = 0.60 to 0.75; OR: 0.70, 95% CI = 0.62 to 0.79; OR: 0.81, 95% CI = 0.72 to 0.91, respectively).

Table 2.

Logistic Regression Models Predicting Recent Successful Smoking Cessation by Self-Reported Local Price

	Bivariate Modela		Adjusted Modelb
	OR (95% CI)	p-Value	OR (95% CI)	p-Value
Local price	1.08 (1.02, 1.15)	.012	1.04 (0.98, 1.11)	.178
Men	0.97 (0.91, 1.03)	.305	0.96 (0.90, 1.01)	.138
Race and ethnicity (white reference)
Non-Hispanic black	0.66 (0.59, 0.74)	<.001	0.78 (0.69, 0.87)	<.001
Hispanic	0.97 (0.85, 1.11)	.676	1.12 (0.98, 1.28)	.096
Non-Hispanic other	0.89 (0.75, 1.04)	.142	0.84 (0.71, 0.98)	.031
Education (college + reference)
Less than high school	0.45 (0.41, 0.50)	<.001	0.52 (0.47, 0.59)	<.001
High school graduate	0.53 (0.49, 0.57)	<.001	0.59 (0.54, 0.64)	<.001
Some college	0.74 (0.69, 0.80)	<.001	0.80 (0.74, 0.86)	<.001
Income ($75 000 + reference)
Less than $15 000	0.55 (0.50, 0.61)	<.001	0.67 (0.60, 0.75)	<.001
$15 000–29 999	0.63 (0.55, 0.71)	<.001	0.70 (0.62, 0.79)	<.001
$30 000–49 999	0.75 (0.66, 0.84)	<.001	0.81 (0.72, 0.91)	.001
$50 000–74 999	0.87 (0.76, 1.00)	.047	0.91 (0.80, 1.04)	.148
N	101 712		101 712

OR = odds ratio. CI = confidence interval.

^aBivariate associations between each individual predictor and smoking cessation with state and year-fixed effects.

^bModel controlled for age, age,² gender, education, CPI-adjusted income, race and ethnicity with smoke-free laws, and state and year-fixed effects.

We show results from models with two-way and three-way interaction terms in Table 3 and Supplementary Appendix Table 2. We did not find any statistically significant effect modification between local price and cessation by race and ethnicity (joint p-value = .637), education (joint p-value = .618), and income (joint p-value = .179), nor did we find any statistically significant interactions in the three-way interaction models exploring interactions between local price, sex, and additional sociodemographic variables, including race and ethnicity (joint p-value = .736), education (joint p-value =.760), and income (joint p-value = .595). For the reader’s convenience, we display the predicted probabilities of cessation associated with each US$1 increment in local cigarette price from interaction models in Supplementary Figures 1–3.

Table 3.

Logistic Regression Models Predicting Recent Successful Smoking Cessation by Self-Reported Local Price with a Two-Way Interaction.

	Race and Ethnicity Interactiona		Education Interactionb		Income Interactionc
	OR (95% CI)	p-Value	OR (95% CI)	p-Value	OR (95% CI)	p-Value
Local price	1.06 (0.99, 1.13)	.097	1.04 (0.97, 1.12)	.254	1.06 (0.98, 1.13)	.137
Race and ethnicity (white reference)
Non-Hispanic black	0.80 (0.59, 1.08)	.149
Hispanic	1.31 (0.97, 1.76)	.075
Non-Hispanic other	0.97 (0.67, 1.42)	.893
Race and ethnicity × price
Non-Hispanic black × price	0.99 (0.92, 1.06)	.805
Hispanic × price	0.96 (0.90,1.03)	.269
Non-Hispanic other × price	0.97 (0.88,1.06)	.445
Education (college + reference)
Less than high school			0.55 (0.41,0.73)	<.001
High school graduate			0.61 (0.49,0.76)	<.001
Some college			0.74 (0.60, 0.91)	.004
Education × price
Less than high school × price			0.99 (0.92, 1.06)	.726
High school graduate × price			0.99 (0.94, 1.04)	.695
Some college × price			1.02 (0.97, 1.07)	.437
Income ($75 000 + reference)
Less than $15 000					0.61 (0.47, 0.80)	<.001
$15 000–29 999					0.68 (0.50, 0.91)	.011
$30 000–49 999					0.92 (0.69, 1.23)	.559
$50 000–74 999					1.12 (0.76, 1.65)	.535
Income × price
Less than $15 000 × price					1.03 (0.97, 1.09)	.398
$15 000–29 999 × price					1.01 (0.95, 1.08)	.701
$30 000–49 999 × price					0.97 (0.91, 1.03)	.274
$50 000–74 999 × price					0.95 (0.88, 1.03)	.214
Joint test of multiplicative interactions (p-value)	.637		.618		.179
N	101 712		101 712		101 712

^aModel included an interaction term between local price and race and ethnicity, adjusting for age, age,² gender, CPI-adjusted income, education with smoke-free laws and state and year-fixed effects.

^bModel included an interaction term between local price and education, adjusting for age, age,² gender, CPI-adjusted income, race and ethnicity with smoke-free laws and state and year-fixed effects.

^cModel included an interaction term between local price and CPI-adjusted income, adjusting for age, age,² gender, education, race and ethnicity with smoke-free laws and state and year-fixed effects.

In sensitivity analyses, we examined models that included additional interactions between sociodemographic characteristics and state and year to control state-specific trends. We observed no statistically significant two-way (Supplementary Appendix Table 3) or three-way interactions (Supplementary Appendix Table 4) between price and sociodemographic characteristics. While the local prices measure likely more closely resemble the prices that smokers encounter in their immediate areas, much of past related literature has relied on state-level average retail prices.⁸ For comparison, we provide results from models that use state retail price as the main predictor (Supplementary Appendix Tables 5–7). We found no statistically significant association between retail price and recent successful cessation with or without effect modification by sociodemographic characteristics. When stratifying our models into pre- and post-federal Family Smoking Prevention and Tobacco Control Act of 2009, we found no statistically significant association in the first period and statistically significant (OR: 1.11, 95% CI = 1.11 to 1.22 joint interaction p = .027) in the second (Supplementary Appendix Table 8). This suggests that local cigarette prices may have become more effective as an aid for smoking cessation over time. When we adjusted for current e-cigarette use in the 2014/2015 and 2018/2019 data, the association between local price and recent successful cessation remained nonsignificant, with the exception of people with less than high school, for whom the odds of cessation increased with higher local prices (OR: 1.39, 95% CI = 1.18 to 1.65; joint interaction p ≤ .001) (Appendices Tables 9–11; Supplementary Figure 4).

Discussion

Our results show that a $1 difference in local price was associated with an approximately 8% difference in the odds of cessation, but only in models not adjusted for individual sociodemographic characteristics. The magnitude of the unadjusted association falls within the range of prior cigarette price elasticity estimates.³⁵ In adjusted models, the association was not statistically significant. The associations between recent successful cessation and other sociodemographic variables closely resembled prior literature: non-Hispanic white adults, adults with higher education, and adults with higher income had greater odds of successful cessation.^34,36 Importantly, in sensitivity analysis, we found that the effectiveness of prices as a tool of tobacco control via cessation may have varied over time. Higher cigarette prices were associated with greater odds of cessation when we only consider the period after 2009. These findings suggest that the dynamic relationship between cessation and local prices may have changed in recent years. We recommend further examination of this possibility by other researchers using the most recent data.

The findings presented in this paper highlight the magnitude of the policy challenge in helping people with established smoking patterns achieve successful cessation. Most adults who smoke, about 68%,³⁷ say they want to quit, and every year, over half of them make a cessation attempt.³⁴ Unfortunately, only about one in eight succeeds.^34,38 People from socially and economically disadvantaged backgrounds are even less likely to successfully quit after an attempt,³⁷ likely because of the greater barriers they face to cessation, such as greater stress and availability of tobacco products.³⁹ While higher cigarette taxes and prices have been linked to lower smoking rates, the research literature, including our own contribution, has provided mixed results regarding whether they improve the odds of cessation in people with established smoking patterns⁴⁰ and how successful they are in promoting cessation among people who smoke and are from socially and economically disadvantaged populations.¹⁴

Cigarette prices in the United States still lag behind many other wealthy nations. For example, the average cost of a pack of the most sold brand cigarettes is more than twice as high in Australia, about two dollars and five dollars more expensive in Canada and the United Kingdom, respectively.⁴¹ In the United States, federal, state, and even local governments, unless barred from doing so by a pre-emption, may impose excise taxes on cigarettes.⁴² The typical prices across the country vary substantially. In some US areas, people who smoke, especially those with limited resources, could be potentially experiencing a large cost burden.⁴³

Given the financial burden of raised prices on people who smoke and are from socially and economically disadvantaged populations, we anticipated their more robust responsiveness to price changes. Our expectation was not born out by the results. We found no statistically significant interactions between local prices by race and ethnicity, education, or income in our main models. However, in models that controlled for e-cigarette use in the last two survey waves when the question was available, we found that people with less than a high school education had greater odds of cessation with increased cigarette prices than college graduates. Our analysis, therefore, provided only suggestive evidence to the literature that argues higher cigarette prices can further health equity by facilitating greater rates of cessation among established smokers from disadvantaged backgrounds. We may speculate that some of the benefits of higher cigarette prices on cessation might be offset by increases in financial stress because of higher cigarette prices, which could prevent successful cessation.³⁹ It is also possible that only the highest levels of cigarette prices were associated with cessation, and the relationship between price and cessation is non-linear. We also cannot rule out other types of changes in smoking behavior, which our successful cessation measure would not capture. For example, related prior work has shown that people who smoke vary in their response to rising prices with respect to smoking intensity, or they may engage in price-minimization strategies, such as purchasing cigarettes in cartons, on American Indian reservations, or over the Internet, using coupons, or switching to cheaper brands.⁴⁴ We recommend a deeper investigation of these dynamics in the future.

Limitations

The findings presented are based on high-quality, detailed, but cross-sectional survey data. Cessation is not directly observed but inferred from reports of recent transition from current smoking to former smoking status. Because prior work showed a great degree of heterogeneity in prices at the sub-state level, likely because of local taxes and price targeting by tobacco retailers, we used a price measure constructed from an average of self-reported prices paid by those within each local area. This measure is endogenous to some extent. For example, tobacco retailers may pass on a tax increase to a lesser degree in local areas where the number of people who smoke is small, and prices are already high or in areas where the ability to absorb higher prices is more limited, such as areas with lower average income, resulting in a lower effect on smoking rates. In contrast, a tax increase in areas with overall low prices of cigarettes may be passed onto consumers fully, or even over-passed as some evidence suggests,⁴⁵ resulting in more cessation. High local prices might also lead to more price-minimizing behavior, such as purchasing cigarettes in cartons or over the Internet. In a sensitivity analysis where we re-estimated all models using state average retail price as the main predictor, we found no associations between price differences across local areas over time and cessation in the fully adjusted models. This suggests that while we consider a local price to be a more meaningful measure of a smoking person’s price environment, our findings are not driven by our choice of a local price measure as opposed to the more commonly used state-level measure. Moreover, the definition of local is not always uniform across respondents. While we used a county or a CBSA as the definition of locality for most, some respondents did not have a county or CBSA marker and were assigned to the “other” category, as detailed in the methods section. Respondents with missing markers were more likely to be from small local areas where few people participated in the survey. The results, therefore, capture the patterns of smaller and more rural areas less well.

Another limitation of this study is that the measure of recent cessation is based on self-reported data, which may not be accurate. A respondent may misremember the date of their last quit attempt. Additionally, respondents, especially those who perceive smoking as socially undesirable, may feel uncomfortable revealing they currently smoke or recently smoked. This would be problematic if such respondents lived in areas where the price of cigarettes is especially high because such a systematic pattern of misreporting would bias our estimates in the positive direction. Respondents may also not have an accurate recollection of how long ago they quit, and some may choose to exaggerate intentionally for the same reasons that some may be reluctant to report they were smokers at all. Studies investigating the discrepancy in smoking status reporting using biomarker data did not find the bias because of under-reporting to be prominent in adults who smoke.⁴⁶ Finally, our respondents who did not self-identify as non-Hispanic white, non-Hispanic black, or Hispanic were grouped into a single category for “other race and ethnicity.” This analytic decision was based on the fact that there were only a small number of sample members in some states who belong to categories outside of non-Hispanic white, non-Hispanic black, and Hispanic. Such grouping leads to an underestimation of heterogeneity within the culturally and socially diverse group, which includes American Indians, Aleuts, Eskimos, people with origins in Asia, Hawaiians and Pacific Islanders, and people who identify with multiple racial or ethnic identities.

Conclusion

Prior research presents strong evidence that raising prices would likely translate into substantial individual and population health benefits: our findings imply that these benefits would be more likely to occur on account of curbed initiation among younger people who do not yet have an established smoking pattern, rather than because of steep increases in cessation among established smokers, although some positive associations between cigarette prices and cessation were found in the late part of the study period. We also do not find disproportionately large benefits for smokers from disadvantaged groups. We encourage further research on the potential of cigarette taxes and prices as a tool to encourage cessation among people with an established smoking pattern.

Supplementary Material

Supplementary material is available at Nicotine and Tobacco Research online.

Declaration of Interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding

Research reported in this publication was supported by the National Cancer Institute (NCI) of the National Institutes of Health (NIMH) [grant number R37CA214787].

Author Contributions

Lucie Kalousova (Conceptualization [equal], Formal analysis [supporting], Investigation [lead], Methodology [lead], Supervision [Supporting], Writing—original draft [Lead], Writing—review & editing [lead]), Yanmei Xie (Formal analysis [lead], Visualization [lead], Writing—original draft [supporting], Writing—review & editing [supporting]), David Levy (Conceptualization [supporting], Funding acquisition [equal], Methodology [supporting], Writing—review & editing [supporting]), Rafael Meza (Conceptualization [supporting], Funding acquisition [equal], Methodology [supporting], Writing—review & editing [supporting]), James Thrasher (Conceptualization [supporting], Funding acquisition [equal], Writing—review & editing [supporting]), Michael Elliott (Formal analysis [supporting], Funding acquisition [equal], Methodology [supporting], Supervision [supporting], Writing—review & editing [supporting]), Andrea R. Titus (Formal analysis [supporting], Writing—review & editing [supporting]), and Nancy Fleischer (Conceptualization [lead], Funding acquisition [lead], Investigation [supporting], Methodology [equal], Project administration [lead], Resources [lead], Supervision [lead], Writing—review & editing [equal]).

Data Availability

The TUS–CPS data are available for download from CPS-IPUMS at https://doi.org/10.18128/D030.V10.0.