Cynthia M. Lakon

Social Disparities in Health: Disproportionate toxicity proximity in minority communities over a decade

This study employs latent trajectory models measuring the level of toxic waste over a decade in the cities of six highly populated, ethnically diverse, counties in southern California from 1990 to 2000 in 3001 tracts. We find that tracts with 15% more Latinos are exposed to 84.3% more toxic waste than an average tract over this time period and tracts with 15% more Asians are exposed to 33.7% more toxic waste. Conversely, tracts with one standard deviation more residents with at least a bachelors degree (15.5%) are exposed to 88.8% less toxic waste than an average tract. We also found that these effects were considerably weaker when using the raw pounds of toxic waste rather than the toxicity-weighted measure, suggesting that future research will want to account for the toxicity of the waste.

Use of propensity score matching in evaluating smokeless tobacco as a gateway to smoking

INTRODUCTION The contentious debate over the promotion of Swedish snus, a form of moist snuff, as an alternative to cigarettes has often focused on the products potential as a gateway to smoking. Findings from prospective studies in the United States have suggested that smokeless tobacco (SLT) is a significant predictor of smoking onset, before and after adjustment for baseline covariates. Use of conventional regression methods in these studies may have resulted in biased parameter estimates, arising from imbalanced covariate distributions in the users and nonusers of SLT. An alternative approach, which has been used widely in the econometric literature, matches exposure or treatment levels on the basis of the propensity score distribution. METHODS Using this approach, we matched current SLT users from the National Longitudinal Study of Adolescent Health with nonusers (496 pairs) and followed them from adolescence into young adulthood for determination of smoking status. RESULTS Prior to matching, the unadjusted risk of becoming a daily smoker was significantly greater for the SLT users compared with nonusers (n = 10,820; range of relative risk = 1.3-2.0, p < .001). However, after pairing individuals on propensity score, we found no evidence for an increased risk of smoking among the SLT users. DISCUSSION Baseline differences in the risk factors for smoking likely account for the association between the two tobacco products.

Network-Based Approaches for Measuring Social Capital

A variety of disciplines, ranging from sociology to public health, have struggled with the conceptualization and measurement of social capital. Differences in the approach used to measure social capital may contribute to variations in the observed relationships between social capital and individual and population health across studies. As well, the heavy reliance on communitarian measures of social capital in public health may truncate the field’s understanding of the relationship between social capital and health (Moore, Shiell, Hawe, & Haines, 2005).

Development of a Twitter-Based Intervention for Smoking Cessation that Encourages High-Quality Social Media Interactions via Automessages

Background The medical field seeks to use social media to deliver health interventions, for example, to provide low-cost, self-directed, online self-help groups. However, engagement in online groups is often low and the informational content may be poor. Objective The specific study aims were to explore if sending automessages to online self-help groups encouraged engagement and to see if overall or specific types of engagement related to abstinence. Methods We conducted a Stage I Early Therapy Development Trial of a novel social media intervention for smoking cessation called Tweet2Quit that was delivered online over closed, 20-person quit-smoking groups on Twitter in 100 days. Social media such as Twitter traditionally involves non-directed peer-to-peer exchanges, but our hybrid social media intervention sought to increase and direct such exchanges by sending out two types of autocommunications daily: (1) an “automessage” that encouraged group discussion on an evidence-based cessation-related or community-building topic, and (2) individualized “autofeedback” to each participant on their past 24-hour tweeting. The intervention was purposefully designed without an expert group facilitator and with full automation to ensure low cost, easy implementation, and broad scalability. This purely Web-based trial examined two online quit-smoking groups with 20 members each. Participants were adult smokers who were interested in quitting and were recruited using Google AdWords. Participants’ tweets were counted and content coded, distinguishing between responses to the intervention’s automessages and spontaneous tweets. In addition, smoking abstinence was assessed at 7 days, 30 days, and 60 days post quit date. Statistical models assessed how tweeting related to abstinence. Results Combining the two groups, 78% (31/40) of the members sent at least one tweet; and on average, each member sent 72 tweets during the 100-day period. The automessage-suggested discussion topics and participants’ responses to those daily automessages were related in terms of their content (r=.75, P=.012). Responses to automessages contributed 22.78% (653/2867) of the total tweets; 77.22% (2214/2867) were spontaneous. Overall tweeting related only marginally to abstinence (OR 1.03, P=.086). However, specific tweet content related to abstinence including tweets about setting of a quit date or use of nicotine patches (OR 1.52, P=.024), countering of roadblocks to quitting (OR 1.76, P=.008) and expressions of confidence about quitting (OR 1.71, SE 0.42, P=.032). Questionable, that is, non-evidence-based, information about quitting did not relate to abstinence (OR 1.12, P=.278). Conclusions A hybrid social media intervention that combines traditional online social support with daily automessages appears to hold promise for smoking cessation. This hybrid approach capitalizes on social media’s spontaneous real-time peer-to-peer exchanges but supplements this with daily automessages that group members respond to, bolstering and sustaining the social network and directing the information content. Highly engaging, this approach should be studied further. Trial Registration Clinicaltrials.gov NCT01602536; https://clinicaltrials.gov/ct2/show/NCT01602536 (Archived by WebCite at http://www.webcitation.org/6WGbt0o1K)

Multiple imputation for missing edge data: A predictive evaluation method with application to Add Health

Recent developments have made model-based imputation of network data feasible in principle, but the extant literature provides few practical examples of its use. In this paper we consider 14 schools from the widely used In-School Survey of Add Health (Harris et al., 2009), applying an ERGM-based estimation and simulation approach to impute the network missing data for each school. Add Healths complex study design leads to multiple types of missingness, and we introduce practical techniques for handing each. We also develop a cross-validation based method - Held-Out Predictive Evaluation (HOPE) - for assessing this approach. Our results suggest that ERGM-based imputation of edge variables is a viable approach to the analysis of complex studies such as Add Health, provided that care is used in understanding and accounting for the study design.

Randomised controlled trial evaluation of Tweet2Quit: a social network quit-smoking intervention

Background We evaluated a novel Twitter-delivered intervention for smoking cessation, Tweet2Quit, which sends daily, automated communications to small, private, self-help groups to encourage high-quality, online, peer-to-peer discussions. Design A 2-group randomised controlled trial assessed the net benefit of adding a Tweet2Quit support group to a usual care control condition of nicotine patches and a cessation website. Participants Participants were 160 smokers (4 cohorts of 40/cohort), aged 18–59 years, who intended to quit smoking, used Facebook daily, texted weekly, and had mobile phones with unlimited texting. Intervention All participants received 56 days of nicotine patches, emails with links to the smokefree.gov cessation website, and instructions to set a quit date within 7 days. Additionally, Tweet2Quit participants were enrolled in 20-person, 100-day Twitter groups, and received daily discussion topics via Twitter, and daily engagement feedback via text. Measures The primary outcome was sustained abstinence at 7, 30 and 60 days post-quit date. Results Participants (mean age 35.7 years, 26.3% male, 31.2% college degree, 88.7% Caucasian) averaged 18.0 (SD=8.2) cigarettes per day and 16.8 (SD=9.8) years of smoking. Participants randomised to Tweet2Quit averaged 58.8 tweets/participant and the average tweeting duration was 47.4 days/participant. Tweet2Quit doubled sustained abstinence out to 60 days follow-up (40.0%, 26/65) versus control (20.0%, 14/70), OR=2.67, CI 1.19 to 5.99, p=0.017. Tweeting via phone predicted tweet volume, and tweet volume predicted sustained abstinence (p<0.001). The daily autocommunications caused tweeting spikes accounting for 24.0% of tweets. Conclusions Tweet2Quit was engaging and doubled sustained abstinence. Its low cost and scalability makes it viable as a global cessation treatment. Trial registration number NCT01602536.

Research Note: The consequences of different methods for handling missing network data in Stochastic Actor Based Models.

Although stochastic actor based models (e.g., as implemented in the SIENA software program) are growing in popularity as a technique for estimating longitudinal network data, a relatively understudied issue is the consequence of missing network data for longitudinal analysis. We explore this issue in our research note by utilizing data from four schools in an existing dataset (the AddHealth dataset) over three time points, assessing the substantive consequences of using four different strategies for addressing missing network data. The results indicate that whereas some measures in such models are estimated relatively robustly regardless of the strategy chosen for addressing missing network data, some of the substantive conclusions will differ based on the missing data strategy chosen. These results have important implications for this burgeoning applied research area, implying that researchers should more carefully consider how they address missing data when estimating such models.

Simulating Dynamic Network Models and Adolescent Smoking: The Impact of Varying Peer Influence and Peer Selection.

We used a stochastic actor-based approach to examine the effect of peer influence and peer selection--the propensity to choose friends who are similar--on smoking among adolescents. Data were collected from 1994 to 1996 from 2 schools involved in the National Longitudinal Study of Adolescent to Adult Health, with respectively 2178 and 976 students, and different levels of smoking. Our experimental manipulations of the peer influence and selection parameters in a simulation strategy indicated that stronger peer influence decreased school-level smoking. In contrast to the assumption that a smoker may induce a nonsmoker to begin smoking, adherence to antismoking norms may result in an adolescent nonsmoker inducing a smoker to stop smoking and reduce school-level smoking.

On social and cognitive influences: relating adolescent networks, generalized expectancies, and adolescent smoking.

We examine the moderating role of friendship and school network characteristics in relationships between 1) youths’ friends smoking behavior and youths’ own generalized expectancies regarding risk and future orientation and 2) generalized expectancies of youths’ friends and youths’ own generalized expectancies. We then relate these constructs to smoking. Using a longitudinal sample from the National Longitudinal Study of Adolescent Health (N = 15,142), the relationship between friends’ generalized expectancies and youths’ expectancies is stronger for those more central in the network, with more reachability, or stronger network ties, and weaker for those with denser friendship networks. Risk expectancies exhibited an inverted U shaped relationship with smoking at the next time point, whereas future orientation expectancies displayed a nonlinear accelerating negative relationship. There was also a feedback effect in which smoking behavior led to higher risk expectancies and lower future orientation expectancies in instrumental variable analyses.

Network Structure, Influence, Selection, and Adolescent Delinquent Behavior Unpacking a Dynamic Process

This study uses National Longitudinal Study of Adolescent Health (Add Health) data to explore the co-evolution of friendship networks and delinquent behaviors. Using a stochastic actor–based (SAB) model, we simultaneously estimate the network structure, influence process, and selection process on adolescents in 12 small schools (N = 1,284) and 1 large school (N = 976) over three time periods. Our results indicate the presence of both selection and influence processes. Moderating effects were tested for density, centrality, and popularity, with only a weak interaction effect for density and influence in the small schools (p < .10). Contexts outside the school affected school networks: adolescents in the large school were particularly likely to form ties to others from equally disadvantaged neighborhoods, and adolescents in the small schools with more outside of school ties increased their delinquency over time. These findings support the importance of delinquency in peer selection and influence processes.