Emily T. Hébert

A Personalized Automated Messaging System to Improve Adherence to Prostate Cancer Screening: Research Protocol

Background Public adherence to cancer screening guidelines is poor. Patient confusion over multiple recommendations and modalities for cancer screening has been found to be a major barrier to screening adherence. Such problems will only increase as screening guidelines and timetables become individualized. Objective We propose to increase compliance with cancer screening through two-way rich media mobile messaging based on personalized risk assessment. Methods We propose to develop and test a product that will store algorithms required to personalize cancer screening in a central database managed by a rule-based workflow engine, and implemented via messaging to the patient’s mobile phone. We will conduct a randomized controlled trial focusing on prostate cancer screening to study the hypothesis that mobile reminders improve adherence to screening guidelines. We will also explore a secondary hypothesis that patients who reply to the messaging reminders are more engaged and at lower risk of non-adherence. We will conduct a randomized controlled trial in a sample of males between 40 and 75 years (eligible for prostate cancer screening) who are willing to receive text messages, email, or automated voice messages. Participants will be recruited from a primary care clinic and asked to schedule prostate cancer screening at the clinic within the next 3 weeks. The intervention group will receive reminders and confirmation communications for making an appointment, keeping the appointment, and reporting the test results back to the investigators. Three outcomes will be evaluated: (1) the proportion of participants who make an appointment with a physician following a mobile message reminder, (2) the proportion of participants who keep the appointment, and (3) the proportion of participants who report the results of the screening (via text or Web). Results This is an ongoing project, supported by by a small business commercialization grant from the National Center for Advancing Translational Sciences of the National Institutes of Health. Conclusions We believe that the use of centralized databases and text messaging could improve adherence with screening guidelines. Furthermore, we anticipate this method of increasing patient engagement could be applied to a broad range of health issues, both inside and outside of the context of cancer. This project will be an important first step in determining the feasibility of personalized text messaging to improve long-term adherence to screening recommendations.

Using Elastic Net Penalized Cox Proportional Hazards Regression to Identify Predictors of Imminent Smoking Lapse

Introduction Machine learning algorithms such as elastic net regression and backward selection provide a unique and powerful approach to model building given a set of psychosocial predictors of smoking lapse measured repeatedly via ecological momentary assessment (EMA). Understanding these predictors may aid in developing interventions for smoking lapse prevention. Methods In a randomized-controlled smoking cessation trial, smartphone-based EMAs were collected from 92 participants following a scheduled quit date. This secondary analysis utilized elastic net-penalized cox proportional hazards regression and model approximation via backward elimination to (1) optimize a predictive model of time to first lapse and (2) simplify that model to its core constituent predictors to maximize parsimony and generalizability. Results Elastic net proportional hazards regression selected 17 of 26 possible predictors from 2065 EMAs to model time to first lapse. The predictors with the highest magnitude regression coefficients were having consumed alcohol in the past hour, being around and interacting with a smoker, and having cigarettes easily available. This model was reduced using backward elimination, retaining five predictors and approximating to 93.9% of model fit. The retained predictors included those mentioned above as well as feeling irritable and being in areas where smoking is either discouraged or allowed (as opposed to not permitted). Conclusions The strongest predictors of smoking lapse were environmental in nature (e.g., being in smoking-permitted areas) as opposed to internal factors such as psychological affect. Interventions may be improved by a renewed focus of interventions on these predictors. Implications The present study demonstrated the utility of machine learning algorithms to optimize the prediction of time to smoking lapse using EMA data. The two models generated by the present analysis found that environmental factors were most strongly related to smoking lapse. The results support the use of machine learning algorithms to investigate intensive longitudinal data, and provide a foundation for the development of highly tailored, just-in-time interventions that can target on multiple antecedents of smoking lapse.

The Time-Varying Relations Between Risk Factors and Smoking Before and After a Quit Attempt

Introduction Intensive longitudinal data (ILD) collected with ecological momentary assessments (EMAs) can provide a rich resource for understanding the relations between risk factors and smoking in the time surrounding a cessation attempt. Methods Participants (N = 142) were smokers seeking treatment at a safety-net hospital smoking cessation clinic who were randomly assigned to receive standard clinic care (ie, counseling and cessation medications) or standard care plus small financial incentives for biochemically confirmed smoking abstinence. Participants completed EMAs via study provided smartphones several times per day for 14 days (1 week prequit through 1 week postquit). EMAs assessed current contextual factors including environmental (eg, easy access to cigarettes, being around others smoking), cognitive (eg, urge to smoke, stress, coping expectancies, cessation motivation, cessation self-efficacy, restlessness), behavioral (ie, recent smoking and alcohol consumption), and affective variables. Temporal relations between risk factors and smoking were assessed using a logistic time-varying effect model. Results Participants were primarily female (57.8%) and Black (71.8%), with an annual household income of <

Concurrent nicotine and tobacco product use among homeless smokers and associations with cigarette dependence and other factors related to quitting

20000 per year (71.8%), who smoked 17.6 cigarettes per day (SD = 8.8). Individuals assigned to the financial incentives group had decreased odds of smoking compared with those assigned to usual care beginning 3 days before the quit attempt and continuing throughout the first week postquit. Environmental, cognitive, affective, and behavioral variables had complex time-varying impacts on smoking before and after the scheduled quit attempt. Conclusions Knowledge of time-varying effects may facilitate the development of interventions that target specific psychosocial and behavioral variables at critical moments in the weeks surrounding a quit attempt. Implications Previous research has examined time-varying relations between smoking and negative affect, urge to smoke, smoking dependence, and certain smoking cessation therapies. We extend this work using ILD of unexplored variables in a socioeconomically disadvantaged sample of smokers seeking cessation treatment. These findings could be used to inform ecological momentary interventions that deliver treatment resources (eg, video- or text-based content) to individuals based upon critical variables surrounding their attempt.

The best of both worlds: Avatar-assisted therapy offers the benefits of therapist-assisted and Internet-based interventions

BACKGROUND Cigarette smoking rates among homeless adults are exceptionally high, contributing to health disparities experienced by this disadvantaged population. Concurrent nicotine and tobacco product use have been shown to result in greater health problems than cigarette smoking alone, and little is known about the rates, motives, and perceived impacts of concurrent use in this group. The purpose of this study is to explore concurrent use rates and constructs of interest among homeless adult daily smokers and to examine differences between concurrent users and non-concurrent users on cigarette dependence, perceived risk of smoking, readiness to quit, and the receipt of recent cessation intervention. METHODS Participants (N = 396) were recruited from six homeless-serving agencies and/or shelters in Oklahoma City. Enrolled participants completed self-report questionnaires. RESULTS The rate of concurrent use was high -67.2%. Participants most frequently endorsed lower cost and a desire to cut down on cigarette smoking as motives for concurrent product use. Concurrent users indicated both a greater likelihood of developing a smoking-related disease if they did not quit for good and a greater number of past year quit attempts relative to non-concurrent users. There was no significant difference between concurrent users and non-concurrent users on readiness to quit or having received recent smoking cessation intervention. CONCLUSION The need for cessation efforts that account for concurrent use for homeless adult smokers is great. Study findings indicate that concurrent users are commonly pursuing the reduction or elimination of cigarette usage and should be specifically targeted for cessation intervention.

The moderating effect of perceived social support on the relation between heaviness of smoking and quit attempts among adult homeless smokers

Avatar-assisted therapy (AAT) for substance use disorders is the focus of an interesting article by Gordon et al. (1), which appears in the current issue of the American Journal of Drug and Alcohol Abuse. In AAT, participants select a digital character that functions as a representative of their identity in an Internet-based virtual treatment environment. A key advantage of AAT is that in-person attendance is not required to receive counseling, which may increase treatment reach and access by attenuating barriers associated with lack of provider proximity and transportation limitations. Studies have shown that certain area-level characteristics such as living in counties with a higher percentage of black, rural, or uninsured residents (2) and greater socioeconomic deprivation (3) are associated with reduced access to substance abuse treatment.Moreover, recent findings from the National Survey on Drug Use and Health have indicated that among individuals who needed drug or alcohol treatment but did not receive it, common reasons for not receiving treatment included: the belief that treatment might have a negative impact on employment (16.1%), transportation limitations or lack of proximal access to preferred treatment (11.8%), inability to find their preferred type of treatment (10.8%), not wanting others to find out that they were receiving treatment (9.6%), and/or the belief that receiving treatment might cause neighbors or communitymembers to have a negative opinion of them (8.3%) (4). AAT offers the ability to address these commonly reported treatment barriers, and reach even the most vulnerable groups. An important feature of AAT is anonymity (5). Group counseling participants can choose to remain anonymous to each other, which may be appealing to individuals who fear the stigma associated withmental health and substance use disorders. Itmay also be possible tomaintain some level of participant anonymity with counselors, even when identifying information is made available to the clinic. Likewise, the selection of avatars that donot alignwith actual physical characteristics, such as skin color, may eliminate the potentially negative impact of implicit bias, which is known to be present among health-care providers (6). The anonymity offered with AAT may help people to feel more comfortable seeking treatment, and treatmentmay be relatively free of biases associated with skin color and other characteristics. Previous intervention research with avatars has focused largely on the treatment of mental health problems (5), and it seems logical to extend this approach to the treatment of substance use disorders. Gordon et al. (1) are among the first to evaluate the feasibility of using AAT to deliver substance abuse treatment. Key study outcomes were rates of treatment completion, drug positive urine screens, and re-arrest during treatment. Nearly half of participants completed the required 16 treatment sessions, which the authors noted is comparable or better than completion rates in other treatment studies (7). Those who completed treatment were less likely to have positive urine screens than non-completers, and no participants were arrested during treatment. Notably, most participants reported that they were interested in AAT because of transportation concerns and interest in technology, with a substantial minority who reported interest due to childcare issues and preference for anonymity. Findings provide initial evidence of feasibility and acceptability, though randomized trials are needed to evaluate the effectiveness of AAT relative to standard treatment approaches. To date, most technology-based interventions for substance use disorders have been automated, self-paced programs such as online educational modules that provide standardized feedback to patients based on their responses (8–10). Automated treatment approaches are effective, convenient, and low cost (i.e., no physical space or treatment provider required) and therefore have the potential for wide reach and impact. However, automated approaches may not be optimally tailored to the needs of the individual, and initial research suggests that having

Financial strain indirectly influences smoking cessation through withdrawal symptom severity

BACKGROUND Over 70% of homeless adults smoke cigarettes. Despite the desire to quit, this group rarely receives the external support to make or maintain a successful quit attempt (SQA; intentional quit attempt lasting >24 h). The Heaviness of Smoking Index (HSI) is a cigarette dependence measure that independently predicts SQAs among domiciled adults. For homeless adults, social support may be a way to buffer the impact of cigarette dependence on SQAs. METHODS The association of the HSI and past-year SQAs, and the potential moderating role of social support, was examined among 445 homeless smokers (Mage = 43.2 + 11.8, 65% male, 57.5% white). Support was measured by the International Support Evaluation List (ISEL-12) and its 3 subscales: tangible, belonging, & appraisal support. RESULTS The HSI was negatively correlated with SQAs (r=-.283, p < .01) and in a regression model controlling for age, sex, and race/ethnicity, appraisal support significantly moderated this relationship (p < .05). The HSI was significantly related to SQAs across low, moderate, and high levels of appraisal support [mean, +1 SD; low (β=-.657, p < .001), medium (β=-.457, p < .001), and high (β=-.258, p < .05)]. Neither the ISEL-12 total nor the other subscales were moderators. CONCLUSION The perceived availability of someone to talk to about ones problems appeared to attenuate the strength of the inverse relationship between the heaviness of smoking and SQAs. Fostering appraisal support for homeless smokers through group treatment may reduce the impact of cigarette dependence on making quit attempts. Social support coupled with the increased availability of empirically-supported cessation aids may improve dismal quit rates among homeless adults.

Introduction to the special issue on use of mobile technology for real-time assessment and treatment of substance-use disorders

BACKGROUND Financial strain has an adverse impact on smoking cessation. However, the mechanisms through which financial strain influences cessation remain unclear. The purpose of the current study was to determine whether financial strain indirectly influenced smoking cessation through withdrawal symptom severity. METHODS Participants (N=139) were primarily Black (63.3%) and female (57.6%) adults enrolled in a smoking cessation program at a safety-net hospital. A self-report financial strain questionnaire was completed one week prior to the scheduled quit date, and the Wisconsin Smoking Withdrawal Scale (WSWS) was completed on the day after the scheduled quit date. Biochemically-verified 7-day point prevalence abstinence was assessed four weeks after the scheduled quit date. Adjusted mediation analyses were conducted using the PROCESS macro for SPSS to evaluate the indirect effects of financial strain on smoking cessation via post-quit withdrawal symptom severity. RESULTS Analyses indicated a significant indirect effect of financial strain on smoking cessation through total withdrawal symptom severity, B=0.027; 95% CI (0.003, 0.066); and specifically anger, B=0.035; 95% CI (0.008, 0.074), anxiety, B=0.021; 95% CI (0.001, 0.051), and sleep symptoms, B=0.015; 95% CI (0.005, 0.043). Greater pre-quit financial strain was associated with greater post-quit withdrawal symptom severity, which increased the likelihood of non-abstinence 4 weeks after the scheduled quit attempt. The direct effect of financial strain on smoking cessation was not significant in any of the mediation models. CONCLUSIONS Findings: suggest that withdrawal severity is an underlying mechanism through which financial strain influences smoking cessation.

Epilogue to the special issue on the use of mobile technology for real-time assessment and treatment of substance use disorders

Until recently, most studies that have examined and intervened upon substance-use disorders have relied upon data collection methods that require multiple in-person visits. In the typical research ...

Understanding the link between contingency management and smoking cessation: The roles of sex and self-efficacy

The current issue of the American Journal of Drug and Alcohol Abuse focuses primarily on the use of mobile technology for real-time assessment of variables relevant to substance use disorders. The articles within characterize predictors of substance use based on data collected in the relevant moment, rather than in retrospect. When ecological momentary assessment (EMA) methods are employed optimally, the timing of variables, such as affect and craving, in relation to substance use can be characterized. There is also the potential to identify bidirectional relationships between variables of interest (e.g., negative affect → relapse → negative affect) and to distinguish within-person versus between-person effects. EMA methods offer the potential for a more detailed understanding of the factors that maintain substance use or lead to relapse among those receiving treatment. Most of the studies in this special issue used EMA to identify predictors of craving, affect, and/or substance use (1–6). The studies focus on opioid (1,6), cigarette (4,7), marijuana (2), alcohol use (3), and general substance use (5). The articles also touch upon other issues related to substance abuse including homelessness (5), pain (4), trauma exposure (2), and post-traumatic stress disorder (PTSD, 3). Kowalczyk et al. (1) reported that among adults with opioid use disorder who were receiving adjunctive clonidine treatment, poorer mood was associated with positive urine screens, whereas there was no relationship between mood and urine screens among participants who received placebo. In addition, cravings for opioids and cocaine were greater when illicit opioids were used, particularly among those who received clonidine. Mood and craving were measured during four random assessments each day over the 12-week intervention period providing detailed information about these key variables. Notably, findings largely reflected within-person relationships between changes in affect and craving, and opioid use. The authors suggested that clonidine may have attenuated the relationship between moderate levels of negative affect and/or craving with substance use. Thus, only more extreme negative affect and craving led to opioid use. Moran et al. (6) characterized sex differences in the causes and contexts of stress and the extent to which stress and drug cues were associated with craving among adults with opioid use disorder. Participants initiated smartphone-based assessments during stress events, and also rated their stress during three random prompts each day. No differences were found in the causes of stress or stress ratings between men and women, and opioid and cocaine cravings increased with greater stress ratings regardless of sex. However, women showed greater increases in opioid craving in the presence of drug cues and when stress was elevated, while men showed greater increases in cocaine craving with increases in stress. Interestingly, the EMA approach allowed the investigators to capture information about companions, location, and activity during stressful events, and they found that women were more likely than men to be with a companion that they have previously used substances with and/or someone who still uses substances. Findings provide useful information that can inform treatment approaches for men and women and, in particular, emphasizes the need for effective strategies to cope with cravings during periods of stress. Previous research has demonstrated strong associations between trauma exposure/PTSD and substance use/disorders (8–10). Two studies in the current issue used EMA to characterize the proximal relations between trauma-related symptoms and substance use. Black et al. (3) examined the temporal associations between PTSD symptoms and subsequent alcohol use