Abstract
Background
Recent increases in violent crime may impact a variety of health outcomes in Mexico. We examined relationships between neighborhood-level violence and smoking behaviors in a cohort of Mexican smokers from 2011–2012, and whether neighborhood-level social cohesion modified these relationships.
Methods
Data were analyzed from adult smokers and recent ex-smokers who participated in Waves 5–6 of the International Tobacco Control Mexico Survey. Self-reported neighborhood violence and social cohesion were asked of Wave 6 survey participants (n=2129 current and former smokers, n=150 neighborhoods). Neighborhood-level averages for violence and social cohesion (range 4–14 and 10–25, respectively) were assigned to individuals. We used generalized estimating equations to determine associations between neighborhood indicators and individual-level smoking intensity, quit behaviors, and relapse.
Results
Higher neighborhood violence was associated with higher smoking intensity (Risk Ratio (RR)=1.17, 95% Confidence Interval (CI) 1.02–1.33), and fewer quit attempts (RR=0.72, 95% CI 0.61–0.85). Neighborhood violence was not associated with successful quitting or relapse. Higher neighborhood social cohesion was associated with more quit attempts and more successful quitting. Neighborhood social cohesion modified the association between neighborhood violence and smoking intensity: in neighborhoods with higher social cohesion, as violence increased, smoking intensity decreased and in neighborhoods with lower social cohesion, as violence increased, so did smoking intensity.
Conclusion
In the context of recent increased violence in Mexico, smokers living in neighborhoods with more violence may smoke more cigarettes per day and make fewer quit attempts than their counterparts in less violent neighborhoods. Neighborhood social cohesion may buffer the impact of violence on smoking intensity.
Keywords: smoking, smoking cessation, violence, social environment, Mexico
Introduction
Violence in Mexico has increased rapidly in recent years, with 32% of households having at least one victim of crime in 2012.[1] Mexican adults rate insecurity and crime as the most important issues affecting their country, above unemployment and poverty.[1] Homicides drastically increased among men in 2008-2009 after a steady decrease from 1990 to 2007 (up to 21.2 deaths per 100,000 men in 2008-2009 from <10 deaths per 100,000 men in 2007),[2] resulting in a decline in life expectancy among 15- to 44-year-old men.[3] Women did not experience a similar drastic increase in homicides.[2] Although drug trafficking in specific states accounts for much of the recent increase in violence, this violence has spread throughout Mexico.[4] Williams proposes two reasons for this situation: (1) social disparities and wealth inequalities are an increasing concern in Mexico, and people with fewer opportunities to participate in the formal economy pursue illegal activities for income; and (2) social bonds may be weakening, reducing Mexican society's ability to regulate individual behaviors, including violence.[5] The impact of violence on public health likely goes beyond its most obvious effects on homicide and victimization.
Smoking may be one health behavior that is influenced by violence. In Mexico, 27% of men and 8% of women were current smokers in 2011, placing it in the middle of adult smoking in the Americas in terms of smoking prevalence.[6] Neighborhoods are important contexts for exploring the impact of violence on smoking and health. For instance, exposure to neighborhood violence or insecurity has been associated with worse mental health outcomes, less physical activity, higher BMI or obesity, and cardiovascular disease outcomes.[7] In addition, these neighborhood factors have been associated with higher smoking prevalence among diverse communities in the US [8-10] and Europe.[11]
These results are consistent with the hypothesis that unsafe neighborhoods with high levels of violence cause stress, which may lead to increases in smoking.[12 13] Other aspects of the neighborhood, though, may buffer these stress pathways, such as neighborhood social cohesion. Indeed, higher neighborhood social cohesion has been associated with lower smoking prevalence in some US communities, [10 14-16] although not all minority communities.[16 17] Other studies have found that collective efficacy, which comprises both social cohesion and social control, are not associated with and smoking prevalence [18] and smoking cessation.[19]
To the best of our knowledge, no studies in low- and middle-income countries (LMICs) have investigated the impact of neighborhood violence and social cohesion on smoking behaviors. In addition, no studies in any country have investigated how neighborhood violence may impact quitting and relapse among smokers. To that end, we investigated the relationship between neighborhood-level violence and smoking intensity, quit behaviors and relapse among a cohort of smokers in Mexico from 2011-2012, and whether the relationships were modified by neighborhood-level social cohesion.
Methods
Population
We data analyzed from the Mexican administration of the International Tobacco Control Policy Evaluation Survey (ITC Mexico), a population based, longitudinal survey of adult smokers in seven major Mexican cities. Data collection started in 2006, and used a stratified, multi-stage sampling scheme with face-to-face interviews. Census tracts were selected from the seven Mexican cities, with probability proportional to the number of households. Two blocks groups were then selected from the census tracts with selection proportional to the number of residents. Households were visited in random order to enumerate household members and recruit eligible participants. Eligible participants were at least 18 years old, had smoked at least once during the previous week, and had smoked at least 100 cigarettes in their lifetimes. Quotas were set for smokers per block and if this number was not reached, another block from the same census tract was randomly selected from which to recruit participants.[20] The same participants were followed from wave to wave, however, due to loss to follow-up, the sample was replenished each year with smokers from the originally selected or adjacent census tracts. The data in this study came from Waves 5 and 6, collected April-May 2011 and October-November 2012, respectively.
We defined three analytic samples of participants: the smoking intensity sample, the quit behavior sample, and the relapse sample. The smoking intensity sample included current smokers at Wave 6 (N=1728; n=1299 followed-up from Wave 5 and n=426 newly recruited at wave 6). The quit behavior sample included participants in Wave 5 who smoked and were present at Wave 6 (n=1384). The relapse sample consisted of participants who had quit at Wave 5 and were present at Wave 6 (n=307). Analytic samples included participants with data on all covariates of interest; n=19 in the smoking intensity sample, n=2 in the quit behavior sample, and n=0 in the relapse sample were excluded due to missing data on the outcome or covariates. Between Waves 5 and 6, 435 of the 1663 participants at Wave 5 were lost to follow-up; n=435 smokers were newly recruited at Wave 6 for replenishment, as described above.
The ITC Mexico Survey was approved by Institutional Review Boards at the Instituto Nacional de Salud Pública (Mexico) and the University of Waterloo (Canada). All participants provided informed consent.
Measures
Smoking intensity
To measure smoking intensity, we compared participants who smoked six or more cigarettes per day to participants who smoked less than six cigarettes per day at Wave 6. Mexican smokers are relatively light smokers, and approximately one-third smoke six or more cigarettes per day.[21]
Smoking cessation
Three dependent variables related to smoking cessation were examined: quit attempts, successful quitting and smoking relapse. A quit attempt was defined for smokers at Wave 5 who indicated at Wave 6 that they had tried to quit smoking in the prior year. A smoker from Wave 5 was considered to have successfully quit if he/she had made a quit attempt in the past year, and had quit for at least one month at Wave 6. A person was considered to have relapsed if they had quit at Wave 5 and reported smoking at Wave 6.
Neighborhood Violence
Neighborhood violence was measured using a four-item index.[22] In Wave 6, participants were asked how often each of the following had occurred in their neighborhood during the past six months: (1) a fight in which a weapon was used, (2) a gang fight, (3) sexual assault or rape, and (4) a robbery or mugging. Possible responses were: never, once, sometimes and many times. Internal consistency was good (α=0.81). Responses were summed for each participant, with higher scores indicating more violence. Each participant was then assigned the average score for all responses in their neighborhood. Data were used from all participants at Wave 6 (i.e., former and current smokers, including newly recruited smokers) who responded to all of these questions (n=2116). We modeled neighborhood violence per the interquartile range (IQR) of 2.71.
Neighborhood Social Cohesion
Neighborhood social cohesion, which is the bond linking a community together, was measured using a five-item index.[22] Residents used a 5-point Likert scale to indicate how strongly they agreed that: (1) people in my neighborhood are nice, (2) people around here are willing to help their neighbors, (3)the people in my neighborhood share strong ties, (4)people in my neighborhood can be trusted, and (5) people in my neighborhood share the same values. Internal consistency was high (α=0.92). Responses were summed for each participant, with higher scores indicating greater social cohesion. Each individual was assigned an average score for their neighborhood, using data from all participants at Wave 6 (i.e., former and current smokers, including newly recruited smokers), who responded to all of these questions (n=2126).
Covariates
Individual socio-demographic variables assessed at each time period included: age, treated as a continuous variable; sex; education categorized as primary education or less, middle school, vocational school/high school/incomplete university, and university/post-graduate; monthly household income, in pesos (approximately $13.24 pesos = USD $1), was categorized as 0-3000, 3001-5000, 5001-8000, ≥8001 and unknown. In a sensitivity analysis we also controlled for neighborhood deprivation, a marker of neighborhood socioeconomic position created for urban census tracts by the National Population Council in Mexico (Consejo Nacional de Población).[23] All covariates included in the models were measured at the wave corresponding to the dependent variable.
Statistical Analysis
Unweighted descriptive statistics were estimated for variables of interest in each analytic sample (i.e., smoking intensity, quit behaviors, and relapse). Using SAS 9.4, generalized estimating equations (GEE) with robust standard errors were used to determine the relationship between neighborhood-level violence and social cohesion, and individual-level smoking and quit behavior outcomes. GEE models allow one to account for the nested design of individuals clustered within neighborhoods. Because the prevalence of the outcomes was higher than 10%, log binomial marginal models were used to estimate the risk ratio (RR). We ran six sets of log binomial models for each outcome studied (i.e., smoking intensity, quit attempts, successful quitting, or relapse). The first model examined the association between neighborhood violence and the outcomes of interest. The second model also included the individual-level socio-demographic covariates. The third model analyzed the bivariate association between neighborhood social cohesion and the outcomes of interest, while the fourth model also adjusted for socio-demographic covariates. The fifth model examined the association between neighborhood violence and neighborhood social cohesion with the outcomes of interest after adjusting for socio-demographic covariates. The final model examined the association between neighborhood violence, neighborhood social cohesion and the interaction between these two variables, with the outcomes of interest, after adjusting for socio-demographic covariates. All models were weighted to account for the sampling design and rescaled to the sample size at the city level to keep the observations from the largest cities from over representing those in smaller cities.
We also conducted several sensitivity analyses. In the first analysis, we adjusted for neighborhood deprivation. Sample sizes were reduced for all analyses including this measure because not all neighborhoods had the variable. Given the relatively high negative correlation between neighborhood violence and neighborhood social cohesion (-0.73), we re-ran our models after excluding observations that, given covariates (including neighborhood social cohesion), had propensities of experiencing levels of neighborhood violence lower than the lowest observed value, and higher than the highest observed value. This analysis was done to ensure our inference used only data that was “on-support,” that is, we did not extrapolate our regression results where there were no data to support our inference, as is sometimes a problem in neighborhood research[24].
Results
Neighborhood violence had a mean between 6 and 7 (range 4–14 in all samples except relapse, which was 4-13), depending on the analytic sample, meaning that, on average, neighborhood residents had seen more than two violent acts in their neighborhoods in the past six months (Table 1). Neighborhood social cohesion had a mean of approximately 18 (range 9.9–24.8), meaning that, on average, neighborhood residents were more likely to agree than disagree with the statements about social bonds in their neighborhoods. One-third of smokers smoked at least 6 cigarettes a day at wave 6 (smoking intensity sample). Approximately 11% of participants had quit for at least one month by wave 6 (quit behavior sample). Nearly 36% had tried to quit within the past year; among those who had tried to quit, 32% were successful. For people who had quit smoking at wave 5, 28% had relapsed by wave 6 (relapse sample).
Table 1. Selected characteristics of study sample, 2011–2012 ITC Mexico Survey.
| Characteristic | Smoking intensity sample N=1728 | Quit behavior sample N=1384 | Relapse sample N=307 |
|---|---|---|---|
| Neighborhood Violence score mean (SD) | 7.0 (2.1) | 6.8 (2.2) | 6.3 (1.8) |
| Range | 4-14 | 4-14 | 4-13 |
| Neighborhood Social Cohesion score mean (SD) | 17.7 (2.8) | 17.7 (2.9) | 17.9 (2.6) |
| Range | 9.9-24.8 | 9.9-24.8 | 9.9-24.8 |
| Age, mean (SD) | 42.7 (15.0) | 43.7 (15.0) | 46.5 (16.4) |
| Sex, n (%) | |||
| Female | 640 (37.0) | 517 (37.4) | 117 (38.1) |
| Education, n (%) | |||
| Primary education or less | 509 (29.5) | 431 (31.1) | 105 (34.2) |
| Middle school | 575 (33.3) | 461 (33.3) | 88 (28.7) |
| Vocational school, high school, incomplete university | 467 (27.0) | 342 (24.7) | 81 (26.4) |
| University and post-graduate | 177 (10.2) | 150 (10.8) | 33 (10.8) |
| Income in pesos, n (%) | |||
| 0-3000 | 421 (24.4) | 349 (25.2) | 90 (29.3) |
| 3001-5000 | 560 (32.4) | 461 (33.3) | 93 (30.3) |
| 5001-8000 | 354 (20.5) | 276 (19.9) | 61 (19.9) |
| ≥8001 | 241 (13.9) | 181 (13.1) | 39 (12.7) |
| Unknown | 152 (8.8) | 117 (8.5) | 24 (7.8) |
| Smoking intensity, n (%) | |||
| Smoking ≥ 6 cigarettes per day | 608 (33.6) | ||
| Smoking <6 cigarettes per day | 1120 (66.4) | ||
| Smoking status, n (%) | |||
| Quit (<30 d) | 4 (0.3) | 2 (0.7) | |
| Quit (>30 d) | 155 (11.2) | 220 (71.7) | |
| Non-daily smoker | 341 (24.6) | 48 (15.6) | |
| Daily smokers | 884 (63.9) | 37 (12.1) | |
| Tried to quit within last year, n (%) | |||
| Yes | 492 (35.6) | ||
| Successful quitting, n (%)a | |||
| Yes | 155 (31.5) | ||
| Relapse, n (%) | |||
| Yes | 85 (27.7) | ||
| People per neighborhood, mean (range) | 12.6 (3-20) | 10.5 (1-19) | 3.2 (1-8) |
| Marginalization index, n (%)b | |||
| Very low | 367 (21.6) | 259 (19.0) | 55 (18.3) |
| Low | 424 (25.3) | 365 (26.8) | 72 (23.9) |
| Medium | 670 (39.5) | 557 (41.0) | 126 (41.9) |
| High/very high | 224 (13.5) | 179 (13.2) | 48 (16.0) |
N=492 in the quit behavior sample
N=1685 in the smoking intensity sample, N=1360 in the quit behavior sample, and N=301 in the relapse sample.
Higher neighborhood violence was associated with a higher intensity of smoking (Table 2; RR=1.17 for a one-IQR increase in neighborhood violence, 95% Confidence Interval (CI) 1.02–1.33 in Model 2; remained significant in other models). Neighborhood social cohesion was not associated with smoking intensity; however, it modified the relationship between violence and smoking intensity (p=0.0007). For smokers living in less cohesive neighborhoods, higher violence was associated with greater risk of high smoking intensity (Figure 1). For smokers living in the more cohesive neighborhoods, higher neighborhood violence was associated with lower smoking. Moreover, for smokers in the less violent neighborhoods, higher social cohesion was associated with a higher prevalence of high smoking intensity, while higher social cohesion was associated with a lower prevalence of high smoking intensity in the most violent neighborhoods (>80%).
Table 2. Adjusted risk ratios of smoking ≥6 cigarettes per day vs. <6 cigarettes per day associated with neighborhood violence and social cohesion, 2012 ITC Mexico Survey.
| Smoking Intensity | ||||||
|---|---|---|---|---|---|---|
| Prevalence ratio (95% CI); n=1728 | ||||||
|
| ||||||
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | |
| Neighborhood Violence (per IQR) | 1.16 (1.02, 1.32) | 1.17 (1.02, 1.33) | 1.20 (1.02, 1.42) | 2.50 (1.59, 3.95) | ||
| Neighborhood Social Cohesion | 0.97 (0.93,1.02) | 0.97 (0.93, 1.01) | 1.01 (0.96, 1.07) | 1.15 (1.05, 1.25) | ||
| Age (in years) | 1.01 (1.00, 1.01) | 1.01 (1.00, 1.01) | 1.01 (1.00, 1.01) | 1.01 (1.00, 1.01) | ||
| Sex | ||||||
| Male | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Female | 0.66 (0.56, 0.79) | 0.67 (0.57, 0.80) | 0.66 (0.56, 0.79) | 0.65 (0.55, 0.78) | ||
| Education | ||||||
| Primary or less | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Middle school | 0.83 (0.66, 1.03) | 0.81 (0.65, 1.01) | 0.83 (0.66, 1.03) | 0.88 (0.71, 1.09) | ||
| Vocational school, high school, incomplete university | 0.72 (0.57, 0.92) | 0.72 (0.57, 0.92) | 0.72 (0.56, 0.92) | 0.77 (0.61, 0.97) | ||
| University and post-graduate | 0.77 (0.56, 1.07) | 0.76 (0.55, 1.04) | 0.77 (0.56, 1.06) | 0.82 (0.59, 1.12) | ||
| Income | ||||||
| 0-3000 | 1.00 | 1.00 | 1.00 | 1.00 | ||
| 3001-5000 | 1.21 (0.96, 1.52) | 1.23 (0.97, 1.56) | 1.21 (0.96, 1.52) | 1.17 (0.95, 1.44) | ||
| 5001-8000 | 1.38 (1.04, 1.82) | 1.42 (1.08, 1.87) | 1.37 (1.04, 1.82) | 1.27 (0.97, 1.67) | ||
| ≥8001 | 1.47 (1.08, 2.00) | 1.47 (1.07, 2.00) | 1.48 (1.09, 2.00) | 1.45 (1.08, 1.94) | ||
| Unknown | 1.36 (0.98, 1.87) | 1.39 (1.01, 1.91) | 1.35 (0.98, 1.86) | 1.29 (0.92, 1.80) | ||
| Violence*Social Cohesion | 0.95 (0.93, 0.98)a | |||||
p-value=0.0007
Figure 1. Predicted probability of smoking 6 or more cigarettes per day for percentiles of neighborhood violence and social cohesion, 2012 ITC Mexico Survey.
Higher neighborhood violence was associated with a lower probability of making a quit attempt within the past year (Table 3; RR=0.72 for a one-IQR increase in neighborhood violence, 95% CI 0.61–0.85 in Model 2; remained significant in other models), while higher neighborhood social cohesion was associated with a higher probability of making a quit attempt (RR=1.04 for a one-unit increase in neighborhood social cohesion, 95% CI 1.00–1.09 in Model 4). The results were attenuated for social cohesion, but not violence, when both neighborhood markers were simultaneously adjusted for (Model 5). No statistical interaction between the two neighborhood markers was found for making a quit attempt. Neighborhood violence was not associated with successful quitting (see Table 3). However, higher neighborhood social cohesion was associated with a higher probability of successful quitting (RR=1.07 for a one-unit increase in neighborhood social cohesion, 95% CI 1.00–1.15 in Model 4). No interaction between neighborhood violence and social cohesion was found for successful quitting.
Table 3. Risk ratios for quit attempts within the last year and successful quitting associated with neighborhood violence and social cohesion, 2011-2012 ITC Mexico Survey.
| Quit Attempts | ||||||
|---|---|---|---|---|---|---|
| Risk ratio (95% CI); n=1384 | ||||||
|
| ||||||
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | |
| Neighborhood Violence (per IQR) | 0.71 (0.60, 0.85) | 0.72 (0.61, 0.85) | 0.64 (0.50, 0.82) | 0.28 (0.09, 0.86) | ||
| Neighborhood Social Cohesion | 1.05 (1.00,1.09) | 1.04 (1.00, 1.09) | 0.96 (0.91, 1.02) | 0.86 (0.74, 1.01) | ||
| Age (in years) | 1.00 (0.99, 1.00) | 1.00 (0.99, 1.00) | 1.00 (0.99, 1.00) | 1.01 (0.99, 1.00) | ||
| Sex | ||||||
| Male | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Female | 1.16 (0.98, 1.38) | 1.17 (0.99, 1.39) | 1.16 (0.98, 1.37) | 1.15 (0.98, 1.36) | ||
| Education | ||||||
| Primary or less | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Middle school | 0.93 (0.75, 1.15) | 0.96 (0.77, 1.20) | 0.93 (0.75, 1.15) | 0.91 (0.73, 1.13) | ||
| Vocational school, high school, incomplete university | 1.04 (0.80, 1.34) | 1.10 (0.85, 1.51) | 1.01 (0.78, 1.31) | 1.99 (0.77, 1.29) | ||
| University and post-graduate | 1.02 (0.75, 1.40) | 1.10 (0.80, 1.53) | 1.01 (0.74, 1.38) | 1.00 (0.74, 1.36) | ||
| Income | ||||||
| 0-3000 | 1.00 | 1.00 | 1.00 | 1.00 | ||
| 3001-5000 | 1.25 (0.99, 1.58) | 1.24 (0.98, 1.57) | 1.27 (1.01, 1.61) | 1.28 (1.01, 1.61) | ||
| 5001-8000 | 1.11 (0.83, 1.49) | 1.13 (0.84, 1.51) | 1.13 (0.84, 1.53) | 1.13 (0.84, 1.53) | ||
| ≥8001 | 1.18 (0.88, 1.60) | 1.17 (0.85, 1.60) | 1.22 (0.91, 1.63) | 1.23 (0.92, 1.63) | ||
| Unknown | 1.16 (0.81, 1.67) | 1.15 (0.80, 1.65) | 1.20 (0.84, 1.71) | 1.17 (0.82, 1.67) | ||
| Violence*Social Cohesion | 1.05 (0.98, 1.13)a | |||||
|
| ||||||
| Successful Quitting | ||||||
| Risk ratio (95% CI); n=492 | ||||||
|
| ||||||
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | |
|
| ||||||
| Neighborhood Violence (per IQR) | 0.85 (0.66, 1.11) | 0.87 (0.66, 1.13) | 1.06 (0.66, 1.47) | 2.24 (0.57, 8.77) | ||
| Neighborhood Social Cohesion | 1.08 (1.01, 1.15) | 1.07 (1.00, 1.15) | 1.08 (0.99, 1.18) | 1.19 (0.97, 1.45) | ||
| Age (in years) | 1.00 (0.99, 1.01) | 1.00 (0.99, 1.01) | 1.00 (0.99, 1.01) | 1.01 (0.99, 1.01) | ||
| Sex | ||||||
| Male | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Female | 1.17 (0.88, 1.57) | 1.16 (0.88, 1.54) | 1.16 (0.86, 1.54) | 1.15 (0.86, 1.53) | ||
| Education | ||||||
| Primary or less | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Middle school | 0.62 (0.41, 0.95) | 0.63 (0.41, 0.96) | 0.63 (0.41, 0.96) | 0.64 (0.42, 0.97) | ||
| Vocational school, high school, incomplete university | 0.62 (0.40, 0.98) | 0.64 (0.41, 0.98) | 0.63 (0.41, 0.98) | 0.63 (0.41, 0.99) | ||
| University and post-graduate | 0.97 (0.58, 1.62) | 0.99 (0.60, 1.64) | 1.00 (0.60, 1.64) | 1.04 (0.64, 1.71) | ||
| Income | ||||||
| 0-3000 | 1.00 | 1.00 | 1.00 | 1.00 | ||
| 3001-5000 | 0.95 (0.66, 1.37) | 0.94 (0.65, 1.36) | 0.94 (0.65, 1.37) | 0.93 (0.65, 1.35) | ||
| 5001-8000 | 0.78 (0.44, 1.38) | 0.78 (0.46, 1.32) | 0.78 (0.46, 1.33) | 0.75 (0.44, 1.27) | ||
| ≥8001 | 0.84 (0.48, 1.46) | 0.84 (0.49, 1.45) | 0.85 (0.49, 2.47) | 0.83 (0.47, 1.46) | ||
| Unknown | 0.53 (0.25, 1.13) | 0.54 (0.25, 1.14) | 0.53 (0.25, 1.13) | 0.53 (0.25, 1.13) | ||
| Violence*Social Cohesion | 0.96 (0.88, 1.03)b | |||||
p-value=0.1312
p-value=0.2675
Neither neighborhood violence nor neighborhood social cohesion was associated with relapse (Table 4). There was also no interaction between the two neighborhood markers and their relationship with relapse.
Table 4. Risk ratios for smoking relapse associated with neighborhood violence and social cohesion, 2011-2012 ITC Mexico Survey.
| Relapse | ||||||
|---|---|---|---|---|---|---|
| Prevalence ratio (95% CI); n=307 | ||||||
|
| ||||||
| Model 1 | Model 2a | Model 3 | Model 4 | Model 5 | Model 6 | |
| Neighborhood Violence (per IQR) | 1.03 (0.83, 1.29) | 1.12 (0.90, 1.40) | 1.29 (0.85, 1.96) | 0.75 (0.25, 2.28) | ||
| Neighborhood Social Cohesion | 1.01 (0.95, 1.07) | 1.00 (0.94, 1.07) | 1.06 (0.93, 1.20) | 0.97 (0.79, 1.21) | ||
| Age (in years) | 0.97 (0.95, 0.98) | 0.97 (0.95, 0.99) | 0.97 (0.95, 0.98) | 0.97 (0.95, 0.98) | ||
| Sex | ||||||
| Male | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Female | 1.12 (0.77, 1.65) | 1.12 (0.76, 1.75) | 1.15 (0.80, 1.67) | 1.13 (0.78, 1.62) | ||
| Education | ||||||
| Primary education or less | 1.00 | 1.00 | 1.00 | 1.00 | ||
| Middle school | 1.28 (0.70, 2.35) | 1.29 (0.70, 2.36) | 1.35 (0.76, 2.42) | 1.40 (0.79, 2.50) | ||
| Vocational school, high school, incomplete university | 1.59 (0.89, 2.82) | 1.53 (0.86, 2.72) | 1.56 (0.87, 2.81) | 1.62 (0.90, 2.94) | ||
| University and post-graduate | 1.95 (1.06, 3.58) | 1.83 (1.01, 3.32) | 1.94 (1.01, 3.74) | 2.06 (1.10, 3.84) | ||
| Income | ||||||
| 0-3000 | 1.00 | 1.00 | 1.00 | 1.00 | ||
| 3001-5000 | 0.91 (0.55, 1.49) | 0.92 (0.55, 1.53) | 0.87 (0.52, 1.47) | 0.84 (0.50, 1.41) | ||
| 5001-8000 | 1.10 (0.68, 1.79) | 1.13 (0.70, 1.82) | 1.09 (0.66, 1.79) | 1.13 (0.69, 1.87) | ||
| ≥8001 | 0.65 (0.29, 1.43) | 0.66 (0.30, 1.45) | 0.65 (0.29, 1.45) | 0.65 (0.29, 1.42) | ||
| Unknown | 0.65 (0.26, 1.62) | 0.65 (0.25, 1.67) | 0.65 (0.26, 1.61) | 0.66 (0.27, 1.62) | ||
| Violence*Social Cohesion | 1.03 (0.97, 1.11)a | |||||
p-value=0.3063
For the sensitivity analyses including neighborhood deprivation, all sample sizes were reduced due to unavailability of the marker for some census tracts (n=1685 for smoking intensity, n=1360 for quit attempts, n=478 for successful quitting, and n=301 for relapse). In all regression models, results were qualitatively similar for all models, with the exception that the interaction between neighborhood violence and social cohesion became statistically significant for the relapse model. Higher neighborhood violence was associated with higher probability of relapse across all levels of social cohesion, but the strength of the relationship was greater as social cohesion increased (p=0.0312). The sensitivity analysis investigating non-positivity also found qualitatively similar results for all models (i.e., the size of the association did not differ greatly from the original analyses and statistical significance remained unchanged). Sample sizes were reduced in the analysis to n=1627 for smoking intensity, n=1297 for quit attempts, n=466 for successful quitting, and n=296 for relapse. All sensitivity analyses available upon request.
Discussion
Among a cohort of smokers in Mexico, higher neighborhood violence was associated with higher smoking intensity and lower probability of making a quit attempt within the past year. Neighborhood violence was not associated with either successful quitting or relapse. The relationship between neighborhood violence and smoking intensity was modified by neighborhood social cohesion; living in a more violent neighborhood was associated with less smoking in high social cohesion neighborhoods, but more smoking in low social cohesion neighborhoods. Neighborhood social cohesion did not modify the relationship between neighborhood violence and quit attempts, successful quitting, or relapse.
Our results for neighborhood markers and smoking intensity are consistent with the limited research on the topic. We found that higher violence was associated with higher smoking intensity. This is consistent with findings, albeit investigating prevalence of smoking in a broader population rather than intensity of smoking among a cohort of smokers, that more violence and feelings of less safety are generally associated with a higher prevalence of smoking.[8-11] Although neighborhood social cohesion was not associated with smoking intensity in our study, it appeared to buffer the impact of violence on smoking intensity. Higher violence was associated with higher smoking intensity in the least socially cohesive neighborhoods, whereas higher violence was associated with lower smoking intensity in the most socially cohesive neighborhoods. Interestingly, in the least violent neighborhoods, higher social cohesion was actually associated with higher intensity of smoking. Although we did not test this directly in our study, it may be possible that social norms help explain this relationship. For instance, one study showed that permissive social norms are more strongly associated with smoking prevalence in neighborhoods with high levels of collective efficacy.[18] In our study, it may be that, in low violence neighborhoods, more cohesive neighborhoods promote higher intensity of smoking through more permissive social norms, whereas in high violence neighborhoods, more cohesive neighborhoods buffer the impact of the stress of those environments, resulting in lower intensity of smoking. Future studies should research these possibilities, particularly in the context of widespread implementation of tobacco control policies that promote the social unacceptability of smoking.[25-27]
Although no other studies have examined the impact of neighborhood violence on cessation behaviors, our finding that living in a more violent neighborhood is associated with a lower probability of making a quit attempt within the past year is consistent with expectations if the neighborhood social environment works through a stress pathway. We did not find a statistically significant relationship between neighborhood violence and successful quitting or relapse. However, relationships were in the expected direction (i.e., higher neighborhood violence associated with less successful quitting but higher probability of relapse) and our small sample (n=492 for successful quitting with n=150 neighborhoods, and n=307 for relapse with n=133 neighborhoods) may have been underpowered. We did find, though, that higher neighborhood social cohesion was associated with more successful quitting. The one other study that investigated collective efficacy and cessation did not find a relationship.[19] These differing results may be due to differences in scales (collective efficacy compared to social cohesion) or environments (New York City versus urban Mexico).
Overall, our findings were consistent with the hypothesis that stress is mediating both the violence-smoking behavior relationships, and the role of social cohesion as a modifier. The hypothesized role of stress in our study is also supported by literature on neighborhood characteristics and stress. For example, in a study among Caucasian and Black women living in Baltimore, researchers found that individuals who perceived more neighborhood problems (e.g., drug and gang activity, street crime) were more likely to experience higher levels of stress.[28] Previous studies also suggest that stressful life events are associated with smoking behavior.[29-31] Hence, stress may be an intermediate variable between violence and smoking behavior. Furthermore, a study conducted among black smokers in Texas found that stress may also be a mediator in the association between social cohesion and smoking cessation.[32] Likewise, neighborhood resources may serve as protective factors against stress. For example, neighborhood features such as faith-based organizations may facilitate the development of social networks, increasing the levels of social support among neighbors and thereby reducing stress levels.[33] Reducing stress levels among neighborhood residents may serve as an important strategy to reduce smoking prevalence and increase the likelihood of successful quitting in Mexico.
Strengths and limitations
To the best of our knowledge, our study is the first to examine the impact of neighborhood violence and neighborhood social cohesion on smoking cessation behaviors. It is also the first in LMICs to investigate the relationship between these two markers of the neighborhood social environment and smoking. There are, however, important study limitations. For instance, all smoking behaviors and assessments of neighborhood social conditions are self-reported and therefore may be biased. However, since we aggregated individual-level responses on neighborhood social conditions to the neighborhood level, we likely eliminated at least some of the potential same-source bias. In addition, self-reported smoking appears reasonably valid in Mexico, as high correlations have been found between salivary cotinine and self-reported cigarette consumption.[34] Our study also suffered some loss to follow-up, which is important for all analyses except the smoking intensity analysis, which is cross-sectional. Between waves 5 and 6, 26% of smokers at wave 5 were lost to follow-up. However, when we compared the distribution of socioeconomic covariates (age, sex, education, income and smoking intensity) in the study sample and among participants who were lost to follow-up, differences were not statistically significant, except that participants with higher education were more likely to be lost to follow-up. In addition, although we tried to model smoking intensity in other ways (e.g., daily light versus non-daily smoking, daily heavy versus non-daily smoking, and a cutoffs of eleven and fifteen cigarettes smoked per day), the regression models would not converge. However, in this cohort, people who smoke six or more cigarettes per day are consistently less like to try to quit and successfully quit compared to people who smoke less often, making this threshold an important cutoff to examine.[21] We did not use continuous cigarettes per day due to the highly skewed, multi-modal nature of the distribution. Finally, we did not investigate the spatial clustering of the outcome or exposure variables due to the limited sample of neighborhoods in the study seven cities; this could be an important future step of this work.
Conclusions
In light of the increased violence in Mexico, smokers living in neighborhoods with more violence may smoke more cigarettes per day and make fewer quit attempts than their counterparts in less violent neighborhoods. Neighborhood social cohesion may buffer the impact of violence on smoking intensity. Our findings indicate that living in neighborhoods with high levels of violence or low levels of social cohesion may be detrimental to non-communicable disease (NCD) risk, including in LMICs. Given the burgeoning global NCD risk, and the high levels of violence in some LMICs, interventions to address the neighborhood social environment may reduce NCD risk globally.
What is already known on this subject?
Neighborhood violence has been associated with worse mental and physical health outcomes, and poor health behaviors, including smoking. However, no studies in low- and middle-income countries (LMICs) have investigated the impact of neighborhood violence on smoking behaviors, and no studies in any country have investigated how neighborhood violence may impact quitting and relapse among smokers. Understanding how the neighborhood social environment influences quitting behavior may pave the way toward novel interventions to reduce smoking.
What this study adds?
We found that in Mexico, a country that has recently experienced an increase in crime and violence, living in a neighborhood with more violence was associated with smoking more cigarettes per day and a lower likelihood of quitting smoking. Living in a more cohesive neighborhood helped buffer some of the impact of smoking more cigarettes per day in more violent neighborhoods. High levels of violence, in addition to the obvious impact on crime and homicide rates, may also impact the growing burden of non-communicable disease risk in LMICs.
Acknowledgments
Funding: Funding for data collection came from the Mexican Consejo Nacional de Ciencia y Tecnología (Salud-2007-C01-70032), with additional funding for analysis provided by the National Cancer Institute at the National Institutes of Health (P01 CA138389) and the Canadian Institutes for Health (57897, 79551, and 115016).
Footnotes
Competing Interests: The authors declare that they have no competing interests.
Contributorship: NLF conceived of the research, supervised the analyses, and drafted and revised the article. PL conducted the analyses and assisted in drafting and revising the article. EAS and LMRS assisted with data interpretation and critically revised the article. JFT conceived of the research, assisted with data interpretation, and critically revised the article. All authors approved the final version for publication.
References
- 1.Encuesta Nacional de Victimización y Percepción Sobre Seguridad Pública 2013 (ENVIPE) Aguascalientes, AGS: Instituto Nacional de Estadística y Geografía; 2013. [Google Scholar]
- 2.Gonzalez-Perez GJ, Vega-Lopez MG, Cabrera-Pivaral CE, Vega-Lopez A, Munoz de la Torre A. Deaths by homicide in Mexico: trends, socio-geographical variations and associated factors. Ciencia & saude coletiva. 2012;17(12):3195–208. doi: 10.1590/s1413-81232012001200005. [DOI] [PubMed] [Google Scholar]
- 3.Gonzalez-Perez GJ, Vega-Lopez MG, Cabrera-Pivaral CE. Impact of homicide on male life expectancy in Mexico. Rev Panam Salud Publica. 2012;32(5):335–42. doi: 10.1590/s1020-49892012001100003. [DOI] [PubMed] [Google Scholar]
- 4.Molzahn C, Ríos V, Shirk DA. Drug Violence in Mexico: Data and Analysis Through 2011. San Diego, CA: Trans-Border Institute, Joan B. Kroc School of Peace Studies, University of San Diego; 2012. [Google Scholar]
- 5.Williams P. El Crimen Organizada y la Violencia en Mexico: Una Perspectiva Comparativa. Revista de Historia International. 2010;11:15–40. [Google Scholar]
- 6.Informe sobre Control del Tabaco para la Región de las Américas. Washington, DC: Organización Panamericana de la Salud; 2013. [Google Scholar]
- 7.Roux AVD, Mair C. Neighborhoods and health. Biology of Disadvantage: Socioeconomic Status and Health. 2010;1186:125–45. doi: 10.1111/j.1749-6632.2009.05333.x. published Online First: Epub Date. [DOI] [Google Scholar]
- 8.Patterson F, Seravalli L, Hanlon A, Nelson DB. Neighborhood safety as a correlate of tobacco use in a sample of urban, pregnant women. Addict Behav. 2012;37(10):1132–37. doi: 10.1016/j.addbeh.2012.05.011. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ganz ML. The relationship between external threats and smoking in Central Harlem. Am J Public Health. 2000;90(3):367–71. doi: 10.2105/Ajph.90.3.367. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Patterson JM, Eberly LE, Ding YM, Hargreaves M. Associations of smoking prevalence with individual and area level social cohesion. Journal of epidemiology and community health. 2004;58(8):692–97. doi: 10.1136/jech.2003.009167. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Miles R. Neighborhood disorder and smoking: Findings of a European urban survey. Social science & medicine. 2006;63(9):2464–75. doi: 10.1016/j.socscimed.2006.06.011. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 12.Ellen IG, Mijanovich T, Dillman KN. Neighborhood effects on health: Exploring the links and assessing the evidence. Journal of Urban Affairs. 2001;23(3-4):391–408. [Google Scholar]
- 13.Feldner MT, Babson KA, Zvolensky MJ. Smoking, traumatic event exposure, and post-traumatic stress: A critical review of the empirical literature. Clin Psychol Rev. 2007;27(1):14–45. doi: 10.1016/j.cpr.2006.08.004. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Echeverria S, Diez-Roux AV, Shea S, Borrell LN, Jackson S. Associations of neighborhood problems and neighborhood social cohesion with mental health and health behaviors: The Multi-Ethnic Study of Atherosclerosis. Health & place. 2008;14(4):853–65. doi: 10.1016/j.healthplace.2008.01.004. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 15.Li SJ, Delva J. Social Capital and Smoking Among Asian American Men: An Exploratory Study. Am J Public Health. 2012;102:S212–S21. doi: 10.2105/Ajph.2011.300442. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kandula NR, Wen M, Jacobs EA, Lauderdale DS. Association Between Neighborhood Context and Smoking Prevalence Among Asian Americans. Am J Public Health. 2009;99(5):885–92. doi: 10.2105/Ajph.2007.131854. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Brown QL, Milam AJ, Smart MJ, et al. Objective and perceived neighborhood characteristics and tobacco use among young adults. Drug Alcohol Depen. 2014;134:370–75. doi: 10.1016/j.drugalcdep.2013.11.005. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ahern J, Galea S, Hubbard A, Syme SL. Neighborhood smoking norms modify the relation between collective efficacy and smoking behavior. Drug Alcohol Depen. 2009;100(1-2):138–45. doi: 10.1016/j.drugalcdep.2008.09.012. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Karasek D, Ahern J, Galea S. Social Norms, Collective Efficacy, and Smoking Cessation in Urban Neighborhoods. Am J Public Health. 2012;102(2):343–51. doi: 10.2105/Ajph.2011.300364. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Thrasher JF, Nayeli Abad-Vivero E, Sebrie EM, et al. Tobacco smoke exposure in public places and workplaces after smoke-free policy implementation: a longitudinal analysis of smoker cohorts in Mexico and Uruguay. Health Policy Plan. 2012 doi: 10.1093/heapol/czs118. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Swayampakala K, Thrasher J, Carpenter MJ, Shigematsu LM, Cupertio AP, Berg CJ. Level of cigarette consumption and quit behavior in a population of low-intensity smokers--longitudinal results from the International Tobacco Control (ITC) survey in Mexico. Addict Behav. 2013;38(4):1958–65. doi: 10.1016/j.addbeh.2012.12.007. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Sampson RJ, Raudenbush SW, Earls F. Neighborhoods and violent crime: A multilevel study of collective efficacy. Science. 1997;277(5328):918–24. doi: 10.1126/science.277.5328.918. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 23.Índice de marginación urbana 2010. México, D.F: Consejo Nacional de Población; 2010. [Google Scholar]
- 24.Ahern J, Cerda M, Lippman SA, Tardiff KJ, Vlahov D, Galea S. Navigating non-positivity in neighbourhood studies: an analysis of collective efficacy and violence. Journal of epidemiology and community health. 2013;67(2):159–65. doi: 10.1136/jech-2012-201317. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Chapman S, Freeman B. Markers of the denormalisation of smoking and the tobacco industry. Tob Control. 2008;17(1):25–31. doi: 10.1136/tc.2007.021386. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 26.Thrasher JF, Perez-Hernandez R, Swayampakala K, Arillo-Santillan E, Bottai M. Policy Support, Norms, and Secondhand Smoke Exposure Before and After Implementation of a Comprehensive Smoke-Free Law in Mexico City. Am J Public Health. 2010;100(9):1789–98. doi: 10.2105/Ajph.2009.180950. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Thrasher JF, Boado M, Sebrie EM, Bianco E. Smoke-free policies and the social acceptability of smoking in Uruguay and Mexico: Findings from the International Tobacco Control Policy Evaluation Project. Nicotine & Tobacco Research. 2009;11(6):591–99. doi: 10.1093/Ntr/Ntp039. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gary TL, Stark SA, LaVeist TA. Neighborhood characteristics and mental health among African Americans and whites living in a racially integrated urban community. Health & place. 2007;13(2):569–75. doi: 10.1016/j.healthplace.2006.06.001. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 29.Matheny KB, Weatherman KE. Predictors of smoking cessation and maintenance. J Clin Psychol. 1998;54(2):223–35. doi: 10.1002/(Sici)1097-4679(199802)54:2<223∷Aid-Jclp12>3.0.Co;2-L. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 30.Parrott AC. Does cigarette smoking cause stress? American Psychologist. 1999;54(10):817–20. doi: 10.1037//0003-066x.54.10.817. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 31.McKee SA, Maciejewski PK, Falba T, Mazure CM. Sex differences in the effects of stressful life events on changes in smoking status. Addiction. 2003;98(6):847–55. doi: 10.1046/j.1360-0443.2003.00408.x. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]
- 32.Reitzel LR, Kendzor DE, Castro Y, et al. The Relation between Social Cohesion and Smoking Cessation among Black Smokers, and the Potential Role of Psychosocial Mediators. Annals of Behavioral Medicine. 2013;45(2):249–57. doi: 10.1007/s12160-012-9438-6. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Stockdale SE, Wells KB, Tang L, Belin TR, Zhang L, Sherbourne CD. The importance of social context: Neighborhood stressors, stress-buffering mechanisms, and alcohol, drug, and mental health disorders. Social science & medicine. 2007;65(9):1867–81. doi: 10.1016/j.socscimed.2007.05.045. published Online First: Epub Date. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Campuzano JC, Hernandez-Avila M, Jaakkola MS, et al. Determinants of salivary cotinine levels among current smokers in Mexico. Nicotine & Tobacco Research. 2004;6(6):997–1008. doi: 10.1080/14622200412331324956. published Online First: Epub Date. [DOI] [PubMed] [Google Scholar]