University of Southern California Dissertations and Theses

Examining the effects of personalized feedback about ALDH2*2, alcohol use, and associated health risks on drinking intentions and consumption: the role of self-efficacy and perceived threat

(USC Thesis Other)

Examining the effects of personalized feedback about ALDH2*2, alcohol use, and associated health risks on drinking intentions and consumption: the role of self-efficacy and perceived threat

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content EXAMINING THE EFFECTS OF PERSONALIZED FEEDBACK ABOUT ALDH2*2, ALCOHOL USE, AND ASSOCIATED HEALTH RISKS ON DRINKING INTENTIONS AND CONSUMPTION: THE ROLE OF SELF-EFFICACY AND PERCEIVED THREAT By Emily B. Saldich A Thesis Presented to the FACULTY OF THE USC DANA AND DAVID DORNSIFE COLLEGE OF LETTERS, ARTS AND SCIENCES UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree MASTER OF SCIENCE (PSYCHOLOGY CLINICAL SCIENCE PROGRAM) December 2024 Copyright 2024 Emily Saldich HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT ii Acknowledgements I want to thank my mentors and committee members, Drs. Susan Luczak, Christopher Beam, and Mark Lai for their support and guidance throughout the course of this project and my graduate career. This study (R21AA028365) was funded by the National Institute on Alcohol Abuse and Alcoholism (NIAAA). HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT iii TABLE OF CONTENTS Acknowledgements..........................................................................................................................ii List of Tables................................................................................................................................... v List of Figures.................................................................................................................................vi Abstract……..................................................................................................................................vii Chapter 1: Introduction....................................................................................................................1 Alcohol Use, Cancer Risk, and ALDH2*2............................................................................... 2 Prevalence of ALDH2*2...........................................................................................................3 Brief Interventions for Alcohol Use in College Students.........................................................4 Brief Interventions with Personalized Genetic and Behavioral Risk Feedback.......................5 Theories of Health-Motivated Behavior Change: Self-Efficacy and Threat Perception..........8 Studies on Self-Efficacy and Threat Perception as Predictors of Health Behavior Change....9 Hypotheses.............................................................................................................................12 Chapter 2: Method.........................................................................................................................13 Participants.............................................................................................................................13 Procedures..............................................................................................................................13 Materials.................................................................................................................................16 Analyses.................................................................................................................................18 Chapter 3: Results..........................................................................................................................24 Descriptive Statistics..............................................................................................................24 Effect of Predictors on Intentions to Change Drinking Behavior..........................................24 Effect of Intentions to Change Drinking Behavior on Change in Heaviest Week Drinking Quantity.................................................................................................................................. 26 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT iv Full Path Model......................................................................................................................27 Chapter 4: Discussion.................................................................................................................... 29 References......................................................................................................................................37 Tables.............................................................................................................................................44 Figures............................................................................................................................................54 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT v LIST OF TABLES Table 1: Study Procedure...............................................................................................................46 Table 2: Means and Standard Deviations of Heaviest Week Drinking Quantity by Wave and Feedback Group with Data Trimmed at 5%..................................................................................47 Table 3: Correlations for Feedback Group, Intentions to Change Drinking Behavior, Heaviest Week Drinking Quantity, Threat Perception, and Self-Efficacy at Waves 2 – 6 Split on the Diagonal by ALDH2*2 Status with Data Trimmed at 5%.............................................................48 Table 4: Robust Multiple Regression Models of the Effect of Predictors on Post-Feedback Intentions to Change Drinking Behavior in the Full Sample and in Drinkers Only, Including Stratified by ALDH2*2..................................................................................................................49 Table 5: Robust Multiple Regression Models of the Effect of Predictors on Post-Feedback Intentions to Change Drinking Behavior in the Full Sample and Split by Low and High Threat Perception...................................................................................................................................... 50 Table 6: Robust Multiple Regression Models of the Effect of Predictors on the Change in Heaviest Week Drinking Quantity from Baseline by Study Wave................................................51 Table 7: Full Path Model Parameter Estimates with Robust Standard Errors...............................54 Table 8: Full Path Model Parameter Estimates for Drinker-Only Sample with Robust Standard Errors..............................................................................................................................................55 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT vi LIST OF FIGURES Figure 1: Multigroup Path Model.................................................................................................. 56 Figure 2: 5% trimmed mean trajectories of heaviest drinking week quantity split by ALDH2*2 status and feedback group..............................................................................................................57 Figure 3: Fitted Multigroup Path Model........................................................................................58 Figure 4: Fitted Multigroup Path Model for the Drinker-Only Sample.........................................59 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT vii Abstract Both flushing and the genetic variant ALDH2*2 are known risk factors for alcohol-related cancers within the research community, but the general population is largely unaware of these risks. This creates potential for brief educational interventions to increase awareness and reduce the alcohol-related cancer burden. Current findings on the degree to which personalized health feedback motivates behavior change are mixed, however. Individual differences like selfefficacy and threat perception are established mechanisms of health behavior change and may explain some of the discrepancies in how people respond to knowledge about health risks. In the current study, we examine the effects of a brief, online feedback intervention about flushing, ALDH2*2, alcohol use, and cancer risk on participants’ intentions to change their drinking behavior and on their actual heaviest week drinking behavior over the following year. We also evaluate the degree to which self-efficacy and threat perception relate to participants’ intentions to change their drinking behavior. Results showed that receiving risk feedback increased participants’ intentions to change their drinking regardless of their ALDH2*2 status. Selfefficacy and threat perception were also related to participants’ intentions to change their drinking behavior. These findings suggest that learning about the cancer risk related to alcohol use affected participants’ intentions to change their drinking behavior, but this, in turn, did not relate to significant changes in their heaviest week drinking quantity in the 10 months postfeedback. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 1 Chapter 1: Introduction Alcohol use has many detrimental effects on health, including increased risk for some cancers. In 2016, the World Health Organization estimated 4.2% of global cancer deaths were attributable to alcohol consumption (World Health Organization, 2018). Certain groups of people are at especially high risk for alcohol-related cancers due to genetic risk factors, such as the variant allele of the alcohol metabolizing gene ALDH2. The variant ALDH2*2 allele is found primarily in people of northeast Asian (Chinese, Japanese, and Korean) ancestry (Eng et al., 2007; Goedde et al., 1992). ALDH2*2 impairs alcohol metabolism through the production of a non-functional aldehyde dehydrogenase enzyme, meaning that when people with ALDH2*2 drink alcohol, a toxic substance called acetaldehyde accumulates in their body (Baan et al., 2007). As a result, people with ALDH2*2 often experience heightened physiological responses to drinking like flushing, i.e. redness in the face, neck, and other parts of the body. Many people are unaware of the association between flushing, ALDH2*2, and cancer risk. Because of this, preventative educational interventions could benefit at-risk individuals, i.e., individuals with ALDH2*2 who drink (Newman et al., 2015). The degree to which genetic feedback motivates changes in substance use behavior is unclear, however (Henrikson et al., 2009). Studies that have implemented genetic feedback interventions with the goal of influencing health-related behaviors like smoking and alcohol use have shown mixed results (Audrain et al., 1997; Hendershot et al., 2010; Lerman et al., 1997; Wright et al., 2006). Because genetic risk information does not always result in behavior change, it is important to understand which mechanisms lead some people to act on personalized risk information while others do not. Theory-driven models of behavior have proposed threat perception (i.e. perceived severity and personal relevance of a health threat) and self-efficacy HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 2 (i.e. perceived ability to respond to a health threat) to be primary drivers of behavior change, such that higher levels of both threat perception and self-efficacy lead to more health promoting behavior (Ajzen, 1985, 1987, 1991; Rimal & Real, 2003; Rogers, 1975; Schwarzer, 2008). These models propose both additive and interactive effects of threat perception and perceived selfefficacy on behavior. This study examines whether a brief, online feedback intervention about the health risks associated with flushing, ALDH2*2, and alcohol use affected participants’ intentions to change their drinking behavior and whether participants’ intentions to change their drinking behavior related to actual changes in their drinking behavior over the course of a year. We also examine the degree to which two proposed mechanisms of behavior change, selfefficacy and threat perception, related to participants’ intentions to change their drinking behavior. Alcohol Use, Cancer Risk, and ALDH2*2 Alcohol consumption is associated with increased risk for oral cavity, pharynx, larynx, esophageal, liver, breast, and colorectal cancers (Baan et al., 2007). Risk is cumulative, such that drinking more alcohol leads to higher risk levels. In a meta-analysis of 572 studies (486,538 cancer cases), heavy drinkers were shown to be 5.13 times more likely to develop oral and pharyngeal cancer and 4.95 times more likely to develop esophageal squamous cell carcinoma compared to non-drinkers (Bagnardi et al., 2015). (Note that individual studies included in this meta-analysis used different ranges to distinguish light, moderate, and heavy drinking. To address this, Bagnardi and colleagues (2015) calculated the midpoint of the range used in each study, and categorized each range as light, moderate, and heavy drinking if the midpoint was less than or equal to 12.5, less than or equal to 50, or greater than 50 grams of alcohol per day, respectively). HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 3 For people with ALDH2*2, the risk for certain alcohol-related cancers is even higher (Chang et al., 2017). Alcohol is primarily metabolized in a two-step process, first by alcohol dehydrogenase (ADH) enzymes, which oxidize alcohol into acetaldehyde, and then by aldehyde dehydrogenase (ALDH) enzymes, which oxidize acetaldehyde into acetate (Li, 2000). The ALDH2 gene that encodes the ALDH2 enzyme is polymorphic: having two common (*1) alleles (ALDH2*1/*1 genotype) results in functional enzyme activity, whereas possessing one variant (*2) allele (ALDH2*1/*2) results in 17-38% of normal ALDH2 enzyme activity and possessing two variant alleles (ALDH2*2/*2) results in 0% of normal ALDH2 enzyme activity (see Chang et al., 2017). As a result, for people with one or two variant ALDH2*2 alleles, drinking alcohol leads to a buildup of acetaldehyde in the body (Baan et al., 2007). This buildup of acetaldehyde causes physiological responses like facial flushing and hives. Because acetaldehyde is a known carcinogen (Baan et al., 2007), it is also associated with increased risk for esophageal and head and neck cancers in those with ALDH2*2 (Chang et al., 2017). A meta-analysis of six studies (945 cancer cases and 2,917 controls) showed heavy drinkers with an ALDH2*2 allele were 3.57 times more likely to develop head and neck cancers than heavy drinkers without ALDH2*2 (Boccia et al., 2009). Similarly, another meta-analysis of 31 studies (8,510 cancer cases and 16,197 controls) found heavy drinkers with an ALDH2*2 allele were 6.50 times more likely to develop esophageal cancer compared to heavy drinkers without ALDH2*2 (Zhao et al., 2015). Because increased cancer risk in those with ALDH2*2 results from deficient alcohol metabolism, only those who drink are at increased risk for these cancers. Additionally, cancer risk increases with the amount of alcohol consumed, meaning heavier drinkers are at higher risk than lighter drinkers (Boccia et al., 2009; Zhao et al., 2015). Prevalence of ALDH2*2 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 4 The ALDH2*2 variant is primarily found in those of northeast Asian ancestry, and is highly prevalent in Chinese, Japanese, and Korean ethnic groups (Goedde et al., 1992). The ALDH2*1/*2 genotype is found in 43% of Chinese American, 40% of Japanese, 32% of Korean American, 30% of Han Chinese/Taiwanese, 26% of Korean, 10% of Thai, and 5% of Chinese/Taiwanese Aborigine samples (Eng et al., 2007). Although the ALDH2*2 allele is only found in certain ethnic groups, these groups make up a large percent of the global population. One research group estimated 8% of the global population has an ALDH2 deficiency (Brooks et al., 2009). Moreover, data from a 2018 World Health Organization report show alcohol use is increasing in the Western Pacific region (World Health Organization, 2018). Brief Interventions for Alcohol Use in College Students College students in the U.S. have high rates of alcohol use and alcohol use disorders (Knight et al., 2002), which has led to the development of interventions to promote healthier alcohol use in college students. Brief preventative interventions have been shown to effectively reduce quantity and frequency of drinking behavior in college students (Carey et al., 2009; Larimer et al., 2004; White et al., 2010), and have been successfully delivered in various modalities, e.g., in person (Marlatt et al., 1998), via mail (S. E. Collins et al., 2002; Larimer et al., 2007), and online (Doumas et al., 2009; Neighbors et al., 2010). Such interventions have often targeted risky drinking behavior in college students by including information on normative peer-group drinking and using motivational interviewing techniques. A meta-analysis of 35 studies (28,621 participants) on the effects of online interventions to reduce college student drinking found that, compared to receiving no intervention, those who received computerdelivered interventions decreased both the quantity and frequency of their alcohol use over the shorter- (≤ 5 weeks) and longer-term (≥ 6 weeks) (Carey et al., 2009). Although effect sizes were HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 5 small in this study, computer-delivered interventions are easily and cheaply disseminated to large audiences and could therefore be useful for promoting healthier alcohol use on a large scale. A more recent meta-analysis of brief motivational interventions in college students, however, suggests results from prior research on this topic should be interpreted cautiously (Huh et al., 2015). This study was an individual participant-level data meta-analysis of brief motivational interventions for alcohol use in college students. Aggregating across 17 studies (8,275 participants), the researchers did not find statistically significant effects of the brief interventions on any drinking in a typical week (Odds Ratio = 0.79, 95% Confidence Interval (CI) = [0.61, 1.10]) or drinking quantity in a typical week (Rate Ratio = 0.96, 95% CI = [0.91, 1.00]). Analyses on shorter- (1-3 months) and longer-term (6-12 months) follow up periods also did not reveal significant intervention effects. Overall, the current literature on brief preventative interventions for alcohol use in college students suggests that, from a public health perspective, these interventions could lead to meaningful change if implemented on a wide enough scale, but that effect sizes are small and additional research is needed to understand which factors underlie intervention efficacy. Brief Interventions with Personalized Genetic and Behavioral Risk Feedback Over the last two decades, brief interventions targeting various health risk behaviors have begun incorporating genetic feedback. Since genome-wide association studies have increased our knowledge about genetic risk factors for disease, researchers have proposed that personalized genetic feedback might enhance preventative interventions by making them more personally salient (Guttmacher & Collins, 2005). On the other hand, genetic factors often only explain a small portion of disease risk (McCarthy & Hirschhorn, 2008), leading to debate about the extent to which genetic risk information motivates behavior change, with many studies showing limited HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 6 or no effects (Henrikson et al., 2009; Hollands et al., 2016; Silarova et al., 2019). There is some evidence, however, that genetic risk information can motivate changes in diet, dental hygiene, and other health-promoting behaviors (Hietaranta-Luoma et al., 2014; Horne et al., 2018; Sparks et al., 2018). Genetic feedback interventions for substance use have mostly been developed for smoking and have shown inconsistent effects on behavior. One study examined how genetic risk feedback altered participants’ smoking cognitions (fear arousal, perceived risks, and benefits of smoking) and behavior (Lerman et al., 1997). Participants received either 1) a minimal-contact quit-smoking counseling session, 2) the counseling session with additional biomarker feedback on alveolar carbon monoxide levels, or 3) the counseling session and biomarker feedback plus genetic feedback on their CYP2D6 genotype and their susceptibility to smoking-related cancer based on their genotype. Although researchers found genetic risk feedback significantly increased perceived risk, quitting benefits, and fear arousal compared to receiving only counseling or counseling with biomarker feedback, they did not find group differences in actual smoking behavior at the 2-month follow-up. After 1 year, the group that received genetic risk feedback did have more quit attempts, although this did not translate into differences in smoking cessation (Audrain et al., 1997). Additional studies have similarly found some short-term but no lasting effects of genetic risk feedback on smoking behavior (Horne et al., 2018; McBride et al., 2002; Sanderson et al., 2008). Some researchers proposed the lack of long-term effects could be due to participants already knowing about the major health risks of smoking prior to their participation (Sanderson et al., 2008). Overall, genetic feedback for smoking has not consistently led to sustained meaningful behavior change. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 7 The association between alcohol use, cancer risk, and ALDH2*2 is not well-known (Newman et al., 2015), however, and educational interventions on this topic may therefore lead to more promising intervention effects than studies on the health risks of smoking. Furthermore, facial flushing serves as a visible reminder of the carcinogenic effects of alcohol, which may help to sustain intervention effects over time. Two studies published to date provided feedback about the health risks associated with ALDH2*2 and alcohol consumption. The first study enrolled 329 Japanese males (Komiya et al., 2006), half of whom were informed about their ALDH2 genotype and the associated health risks (alcohol use problems and liver dysfunction for ALDH2*1/*1, cancer risk for ALDH2*1/*2, and acute alcohol problems for ALDH2*2/*2). Participants in this study were employees at a manufacturing company and most of the participants were in their 30s and 40s. Drinking frequency and liver function were measured 7 months prior to the intervention and 18 months after the intervention. Komiya and colleagues did not find significant changes in participants’ drinking frequency or liver function pre- to postfeedback for those with or without an ALDH2*2 allele, although they do report a non-significant trend of increased drinking frequency among those without ALDH2*2 and among those with ALDH2*2 who were not notified of their genotype, and decreased drinking frequency among those with ALDH2*2 who were notified of their genotype. The second study on alcohol use and ALDH2*2 by Hendershot et al. (2010) enrolled 200 Asian-American college students (53.5% female) and assigned half to a risk feedback group and half to a control group. They informed feedback group participants that the ALDH2 genotype affects the functioning of alcohol metabolizing enzymes, and that ALDH2*2 is associated with increased acetaldehyde buildup and decreased rates of alcohol dependence. Participants then viewed their personal genotype and associated risk factors (increased risk for alcohol dependence HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 8 for ALDH2*1/*1, increased cancer risk for ALDH2*1/*2, and decreased risk for alcohol dependence for ALDH2*2/*2). The researchers examined alcohol use at baseline and at 30 days post-feedback and found significant intervention effects: participants with ALDH2*2 who received risk feedback had greater reductions in their drinking frequency and typical drinking quantity after 30 days compared to those with ALDH2*2 who did not receive risk feedback. In summary, there is some evidence that genetic feedback can motivate participants to change their cognitions about substance use (e.g., perceived risk from use, intentions to quit using) and their actual substance use behavior. Interventions on the ALDH2-drinking-cancer risk relationship are underexplored, and, unlike genetic-smoking-cancer feedback interventions, provide novel information about cancer risk. Further research is needed to understand which factors lead people to act on genetic feedback interventions so that behavior change can be maximized. Theories of Health-Motivated Behavior Change: Self-Efficacy and Threat Perception Models including Protection Motivation Theory (Rogers, 1975), the Theory of Planned Behavior (Ajzen, 1985, 1987, 1991), the Risk Perception Attitude Framework (Rimal & Real, 2003), the Extended Parallel Process Model (Witte, 1992), and the Health Action Process Approach (Schwarzer, 2008), among others, incorporate factors related to threat perception and self-efficacy as predictors of health-related intentions and behaviors. Threat perception is the perceived severity and personal relevance of a health risk. It has been theorized that people need to feel sufficiently threatened by a health risk to act on it. Similarly, self-efficacy, or a person’s perceived control over responding to a health risk, has also been theorized to motivate behavior change since people are unlikely to change their behavior if they do not believe this will alter their probability of experiencing health consequences. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 9 The primary and proximal predictor of behavior in these models is the intention to perform a behavior, for example, a person’s intentions to drink alcohol is the primary predictor of their actual alcohol use. Threat perception and self-efficacy are proposed to predict either intentions, or intentions and behavior, as in the Health Action Process Approach model (Schwarzer & Fuchs, 1995). Overall, health behavior models consistently hypothesize that in order for behavior change to take place, one must both feel sufficiently threatened by a health risk and also feel some degree of control over reducing their risk. Some models propose an additive effect such that self-efficacy and threat perception contribute independently to health intentions (Witte, 1992), whereas others propose an interaction between the two constructs (Rogers, 1975), such that a certain level of threat perception is required for self-efficacy to have an effect. Note that each of these models uses different terms related to self-efficacy, including perceived efficacy, self-efficacy, perceived behavioral control, and response efficacy. When describing prior literature, I adopt the terms used in the studies when describing their results. I use the term self-efficacy when referring to the variable we obtained in this study. Studies on Self-Efficacy and Threat Perception as Predictors of Health Behavior Change Studies testing behavioral motivation models support the relevance of threat perception and self-efficacy for shaping health intentions and behaviors related to drug and alcohol use, cancer screening, dental hygiene, diet, and food safety, among others (Armitage et al., 1999; Armitage & Conner, 1999; McCaul et al., 1993; Mcmillan & Conner, 2003; Milton & Mullan, 2012; Schnoll et al., 2005; Wright et al., 2006). A meta-analysis by Witte and Allen (2000) of 98 studies examined how altering the content of threatening messages affects attitudes, intentions, and behavior change. They found main effects suggesting that increasing both the threat content (higher degree of fear appeal, severity, and susceptibility) and the efficacy content (higher degree HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 10 of self-efficacy and response-efficacy) of a message led to significant change in attitudes, intentions, and behavior. They also examined interaction effects by creating four message groups: high threat/high efficacy, high threat/low efficacy, low threat/high efficacy, and low threat/low efficacy. They found low threat messages had the least impact on attitudes, intentions, and behaviors regardless of efficacy content, while high threat/high efficacy had the greatest impact. Levels of threat perception and self-efficacy have also been shown to affect substance use cognitions and behavior. A study by Armitage et al. (1999) applied the Theory of Planned Behavior to alcohol and cannabis use and found measures of efficacy (self-efficacy and perceived behavioral control) predicted intentions to drink alcohol, and intentions to drink was the only significant predictor of alcohol use. Similarly, both self-efficacy and perceived behavioral control were significant predictors of intentions to use cannabis, and intentions to use cannabis predicted actual cannabis use. Conner et al. (1999) also found support for the Theory of Planned Behavior for alcohol use, with perceived behavioral control (likened to self-efficacy) predicting intentions to drink alcohol and both perceived behavioral control and intentions to drink alcohol predicting past two-week alcohol use frequency. Additional studies have specifically examined how threat perception and self-efficacy affect behavior in the context of providing participants with personalized feedback about health risks. Scholl et al. (2005) provided participants with personalized cancer risk feedback based on factors like family history, smoking, diet, and exercise. Results indicated both quitting selfefficacy and perceived cancer risk significantly predicted smoking status and readiness to quit smoking 2 years after baseline measurements were obtained. Wright et al. (2006) studied how feedback about genetic risk for coronary heart disease affected smoking intentions, showing that HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 11 those who received risk information felt more susceptible to coronary heart disease (threat perception) and had higher intentions to quit smoking compared with the control group. They also found that self-efficacy was associated with greater intentions to quit smoking across groups. Taken together, personalized genetic and behavioral risk feedback has affected healthrelated cognitions and behavior in some cases, but not others. Additional studies have found threat perception and self-efficacy predict change in substance use and other health behaviors, suggesting these factors may be important mechanisms of health-related behavior change. In the current study, participants in the intervention group received information about the associations between flushing, the ALDH2*2 allele, drinking alcohol, and cancer risk, after which their alcohol use and other relevant variables were tracked for one year. This study builds upon the two previous brief genetic feedback studies for ALDH2*2 and alcohol use (Hendershot et al., 2010; Komiya et al., 2006). Our study differs in that it includes 6 waves of data collection, allowing for the examination of alcohol use trajectories over time. Additionally, we had two risk feedback conditions: one in which participants receive feedback about health risks associated with their genotype as in Hendershot et al. (2010) and Komiya et al. (2006), and one in which they only receive risk feedback about their phenotype (flushing). The addition of the phenotype feedback condition allows us to determine whether genetic feedback is necessary to motivate behavior change, or whether feedback about flushing alone is enough to promote behavior change. If this is the case, dissemination of this feedback intervention would be feasible on a much wider scale. In addition to examining overall intervention effects on participants’ intentions to change their drinking and their actual drinking behavior, the current study will examine the degree to which self-efficacy and threat perception are associated with participants’ HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 12 intentions to change their drinking and the degree to which participants’ intentions to change their drinking predict a change in their actual drinking behavior. Hypotheses Hypothesis 1: Among drinkers with ALDH2*2, those who receive risk feedback will have higher intentions to change their drinking behavior than those who do not receive risk feedback. Hypothesis 2: Among drinkers with ALDH2*2, we expect higher post-feedback threat perception, self-efficacy, and their interaction to be associated with greater intentions to change drinking behavior. We will explore whether these are additive and/or interactive effects. We will also explore whether b) risk feedback will mediate the relationship between a) baseline threat perception, self-efficacy, and their interaction and c) intentions to change drinking behavior. Hypothesis 3: Higher intentions to change drinking will be associated with decreases in heaviest week drinking quantity over the course of the study. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 13 Chapter 2: Method Participants Undergraduates of Chinese, Korean, Japanese, or Vietnamese ancestry were recruited to participate in this study (58% female; age range 17-25). Two cohorts of students were enrolled for a total of 360 participants. The first cohort included incoming students beginning their first year of college (n = 177), and the second cohort included students who had completed at least one year of college and also endorsed flushing from alcohol (n = 183). Procedures Recruitment The incoming student cohort was recruited through an online training that was required before arriving on campus and that taught students about health and safety related to alcohol use (AlcoholEdu® for College). At the end of the training, they were shown a link to the study opportunity where they could indicate their interest in participating. Once classes began, they were also recruited by flyers posted around campus that stated, “USC undergraduates 17-25 years old of Chinese, Japanese, Korean, or Vietnamese heritage wanted as participants for a research study.” Participants emailed the laboratory about their interest in participating. The flushing cohort was recruited via the same flyers. After indicating their interest, participants completed a 6-item screener survey that obtained age, biological sex, ethnicity, semester they started at any college, semester they started at USC, and expected graduation semester. Students who did not identify themselves as incoming students also reported whether or not they flushed from one or two drinks of alcohol when they first started drinking. Study Procedure HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 14 See Table 1 for a summary of the study procedure. Six waves of data collection were completed over the course of ~15 months. For wave 1, all 360 participants were brought into the laboratory for an initial 1-hour session in which they gave their written informed consent, completed a battery of online surveys, and provided a saliva sample for genotyping. Before wave 2, participants were genotyped and then randomized based on ALDH2 genotype and lifetime drinking status into three groups to view one of the three feedback sessions described below. All other sessions were completed entirely online. At wave 2, a link was emailed to participants, which took them through an hour-long session. During the session, participants completed a prefeedback survey, viewed their assigned feedback session, and completed a post-feedback survey. Directly following the feedback session, participants reported on their intentions to change their drinking behavior as a result of this study and filled out a brief comprehension quiz. Participants completed four additional rounds of repeated measures online surveys 1-, 4-, 7-, and 10-months post-feedback. At all six time points, participants self-reported their recent alcohol consumption, cancer-specific self-efficacy and cancer-specific threat perception, among other measures not used in this study. Feedback Sessions All three online feedback sessions were designed to take approximately 15-20 minutes to complete. Participants clicked through a series of slides with images, text, and animations. Each slide also had an accompanying voiceover, and participants were required to remain on each page until the voiceover finished playing. The flushing phenotype (PHEN) feedback session included information on alcohol use and cancer risk, as well as information on how flushing and ALDH2*2 affect risk. First, participants were informed that drinking alcohol increases risk for oral cavity, pharynx, larynx, HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 15 esophageal, liver, breast, and colorectal cancers. Next, participants learned that cancer risk increases even further for those who flush from alcohol. Participants viewed information about alcohol metabolism, the association between flushing and the ALDH2 gene, and the prevalence of ALDH2*2 for various Asian ethnic groups. They also learned that the ALDH2*1/*2 and ALDH2*2/*2 genotypes are risk factors for alcohol-related cancers. Lastly, participants read about ways to reduce alcohol-related cancer risk like reducing alcohol consumption and having good oral hygiene. Participants in the flushing phenotype plus genetic feedback (PHEN+GENE) group viewed the PHEN feedback session, plus two additional slides informing them of their own ALDH2 genotype and the specific health risks associated with their genotype (alcohol-related cancer for ALDH2*1/*2 and ALDH2*2/*2 and alcohol dependence for ALDH2*1/*1). Thus, all participants in the PHEN or PHEN+GENE groups got some degree of risk feedback, both about the general association between alcohol use and cancer risk regardless of genotype, and about the specific health risks associated with all three ALDH2 genotypes. The attention control (CONTROL) feedback was designed to include similar topics to the other feedback sessions relating to being a USC student, ethnicity, alcohol use, and genetic information. First, participants viewed demographic data on race/ethnicity and international student country of origin across three USC classes. Next, participants viewed alcohol use results from the AlcoholEdu training program (e.g., where USC students typically drink on campus and top reasons why students drink or limit their drinking stratified by Asian and non-Asian USC undergraduates). None of the alcohol-related data provided normative feedback about typical quantity or frequency of USC student alcohol use. The CONTROL session also included comparisons of substance use across academic majors and across two personality dimensions HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 16 (harm avoidance and novelty seeking). Lastly, participants read about another gene, its associations with personality and alcohol use, and the prevalence of genotypes across ethnic groups. Materials Alcohol Consumption Measures Alcohol outcomes were adapted from the Daily Drinking Questionnaire (DDQ; Collins et al., 1985), a self-report questionnaire that includes items on alcohol use frequency and quantity. Participants were informed at the beginning of the questionnaire that “One standard drink is a 12 oz bottle of beer, 5 oz glass of wine, or 1.5 oz shot of liquor or mixed drink with one shot of liquor.” Participants reported the number of standard drinks they had each day of the week during a typical drinking week and during their heaviest drinking week in the past 30- or 90-day period. Participants also reported the frequency of drinking any alcohol, drinking 4+ (for females) or 5+ (for males) drinks, and drinking 10+ (for females) or 12+ (for males) drinks in the past 30- or 90-day period. The alcohol use frequency items were scored on an 8-point graduated frequency scale for 30-day drinking, with response options ranging from 1 (“Never”) to 8 (“Daily/Almost Daily”), and on a 10-point graduated frequency scale for 90-day drinking, with response options ranging from 1 (“Never”) to 10 (“Daily/Almost Daily”). Lastly, participants reported the maximum number of standard drinks they consumed within the 30- or 90-day period and the amount of time it took to drink them. We chose the total number of standard drinks consumed in participants’ past-30-day heaviest drinking week as our primary alcohol use outcome to capture more variability in participants’ drinking behavior. We refer to this variable as heaviest week drinking quantity throughout the study. The heaviest week drinking quantity variable was highly skewed in the HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 17 positive direction, with most participants reporting no or minimal drinking during their heaviest drinking week at each wave. To create a more meaningful variable, we calculated difference scores that represent the change in participants’ heaviest week drinking quantity from baseline to 1-, 4-, 7-, and 10-months post-feedback, creating a more symmetrical outcome variable. For example, we subtracted wave 2 heaviest week drinking quantity from wave 3 heaviest week drinking quantity to obtain a difference score representing participants change in drinking at 1- month post-feedback. Disease-Specific Self-Efficacy This survey was adapted from a risk perception survey for diabetes (Walker et al., 2003) to measure cancer-specific self-efficacy and threat perception. The survey includes 8 items divided into 3 subscales, including the 4-item personal control/self-efficacy subscale that measures the amount of control participants feel they have over developing cancer. This 4-item subscale is comprised of two sub-domains of self-efficacy: external control (e.g., “If I am going to get cancer, there is not much I can do about it”), and internal control (e.g., “I think that my personal efforts will help control my risks of getting cancer.” The second subscale is a 2-item optimistic bias/comparative risk subscale that measures participants’ perception of their own risk for developing cancer compared to other people (e.g., “Compared with other people of my same age and sex, I am less likely than they are to get cancer”). The third is a 2-item worry subscale that measures participants level of worry about getting cancer (e.g., “I worry about getting cancer”), and was not collected in the current study. The survey is scored on a 4-point Likert scale, with response options ranging from “strongly agree to “strongly disagree.” Survey items were coded such that higher values indicate higher levels of self-efficacy and comparative risk (Rochefort et al., 2020; Walker et al., 2003). For this study we initially chose the 4-item personal HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 18 control/self-efficacy subscale as our measure of self-efficacy, however Cronbach’s alpha was low for this subscale (α = 0.64 at wave 3), prompting us to use the more reliable external control sub-domain instead (α = 0.76 at wave 3). This variable was obtained at all study waves (prefeedback at wave 2; see Table 1). Threat Perception To measure participants’ perception of their future cancer risk, we use a single item: “When I get older, I will be more likely than others my same age to get cancer.” The survey is scored on a 4-point Likert scale, with response options ranging from “strongly agree to “strongly disagree,” and was coded such that higher values correspond with higher threat perception. This item was used to capture threat perception in the analyses below. This variable was obtained at all study waves (pre-feedback at wave 2; see Table 1). Intentions to Change Drinking Participants’ intentions to change their drinking behavior was measured with a single item: “As a result of taking this training, how much will your drinking behavior change?” The item has six response options: 1 = Not at all, 2 = Slightly, 3 = Somewhat, 4 = Moderately, 5 = Fairly much, 6 = A great deal). This variable was obtained post-feedback at wave 2 (see Table 1). Analyses Data were analyzed in the programming language R (version 4.3.0; R Core Team, 2022) with the dplyr (version 1.1.2; Wickham et al., 2023), psych (version 2.3.6; Revelle, 2023), MASS (version 7.3.58.4; Venables & Ripley, 2002), lme4 (version 1.1.33; Bates et al., 2015), tidyr (version 1.3.0; Wikham et al., 2023), OpenMx (version 2.21.8; Baker et al., 2023), semTools (version 0.5-6; Jorgensen et al., 2022), nnet (version 7.3.18; Venables & Ripley, HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 19 2002), usdm (version 2.1.7; Naimi et al., 2014), and devtools (version 2.4.5; Wickham et al., 2022) packages. Graphs were created using the ggplot2 (version 3.4.2; Wickham, 2016) package. Analyses included data from wave 2 (pre-feedback) through wave 6 (10-months postfeedback). Drinking data from wave 1 were not included as this wave was collected over 3 months of the fall semester so the timeframes for which participants reported their past-30- and 90-day drinking varied too greatly across participants. The first step of our analyses was to examine the descriptive statistics for all our variables, including means of heaviest week drinking quantity, intentions to change drinking behavior, self-efficacy, and threat perception, as well as the correlations of all variables across study waves. Polychoric correlations were used for correlations between two ordinal variables with fewer than eight categories, whereas Pearson correlations were used for variables with more than eight categories. We then used a series of analytic approaches to evaluate the degree to which our predictor variables of interest (feedback group, ALDH2*2 status, self-efficacy, and threat perception) interacted to predict participants’ intentions to change their drinking behavior, as well as the degree to which participants’ intentions to change their drinking behavior affected their heaviest week drinking quantity. Multiple Regression Models In the first approach, we used multiple regression models to evaluate effects of the proposed predictors on participants’ intentions to change their drinking behavior. Specifically, we examined whether ALDH2*2 status (X1i), feedback group (X2i), the interaction between ALDH2*2 status and feedback group (X1iX2i), self-efficacy (X3i), threat perception (X4i), and the interaction between self-efficacy and threat perception (X3iX4i) were significantly associated with participants’ intentions to change their drinking behavior (Yi) post-feedback. For all HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 20 multiple regression models, we used self-efficacy and threat perception obtained at wave 3 (1- month post-feedback). We also included baseline heaviest week drinking quantity (X5i) as a predictor in the model to account for the likelihood that those who drank more alcohol prior to receiving the feedback may have been more motivated to reduce their drinking after learning they are at risk, compared with those who drank less or not at all and may not see a need to change their drinking. We also covaried for age (X6i) and sex (X7i). Yi = β0 + β1*X1i + β2*X2i + β3*X1iX2i + β4*X3i + β5*X4i + β6*X3iX4i + β7*X5i + β8*X6i + β9*X7i + εi. We first ran the model on the full sample, then split the sample by ALDH2*2 status to examine these relationships among those with and without ALDH2*2, as we would expect stronger effects among participants who learned they are at increased risk for alcohol-related cancers based on their genotype: Yi = β0 + β1*X2i + β2*X3i + β3*X4i + β4*X3iX4i + β5*X5i + β8*X6i + β9*X7i + εi. Similarly, because we only expected risk feedback to be associated with greater intentions to change drinking behavior among participants who consume alcohol, we ran these models again in the drinker-only sample, excluding participants who reported zero drinks during their heaviest drinking week at all five study waves. Because the data were right-skewed, we also ran the model using a robust regression estimator to account for violations of model assumptions. The robust regression models produced the same pattern of results as the traditional models, and we present the results from the robust regression models here to match our other analyses. In our next set of analyses, we used multiple regression models to examine whether the risk feedback and participants’ intentions to change their drinking behavior were associated with changes in their heaviest week drinking quantity from baseline to 1-month post-feedback. To HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 21 obtain a difference score representing the change in heaviest week drinking quantity over the month following the feedback, we subtracted baseline heaviest week drinking quantity (at wave 2) from post-feedback heaviest week drinking quantity (at wave 3), with positive difference scores indicating increased heaviest week drinking, and negative difference scores indicating decreased heaviest week drinking. After plotting the distribution of the resulting difference scores, we determined that the overall pattern of the data was normally distributed, although we did observe many zeroes indicating no change in heaviest week drinking quantity as well as some extreme outliers in the data. We therefore determined that these data did not meet the model’s normality assumption and opted to use a robust regression model that would be less sensitive to the influence of extreme outliers and produce more reliable parameter estimates. In this set of analyses, we examined whether ALDH2*2 status (X1i), feedback group (X2i), the interaction between ALDH2*2 status and feedback group (X1iX2i), intentions to change drinking behavior (X3i), and baseline heaviest week drinking quantity (X4i) significantly predicted change in heaviest week drinking quantity (Yi) over this month-long period. We again included age (X5i) and sex (X6i) as covariates. Yi = β0 + β1*X1i + β2*X2i + β3*X1iX2i + β4*X3i + β5*X4i + β6*X5i + β7*X6i + εi. Then, to determine whether changes in heaviest week drinking quantity persisted past the 30-day post-feedback period, we repeated the multiple regression analyses for the remaining waves of the study. To do this we subtracted baseline heaviest week drinking quantity at wave 2 from heaviest week drinking quantity at waves 3, 4, 5, and 6, obtaining difference scores that represent longer-term changes in heaviest week drinking quantity. We used the resulting difference scores as the outcome variables in a series of runs of the model described above. We then split the HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 22 sample by ALDH2*2 status to examine these relationships among those with and without ALDH2*2 at all study waves. Yi = β0 + β1*X2i + β2*X3i + β3*X4i+ β6*X5i + β7*X6i + εi. We repeated these analyses in the drinker-only sample. Hochberg’s method was used to adjust for multiple comparisons (Hochberg, 1988). Path Models Our final step was to include all variables in a full multigroup path model (depicted in Figure 1) to further explore these relationships grouping participants by their ALDH2*2 status. Our full path model includes intentions to change drinking behavior as the primary proximal predictor of change in drinking behavior (change in baseline to 1-month post-feedback heaviest week drinking quantity), and the feedback group (PHEN and PHEN+GENE) mediating the effects of self-efficacy and threat perception on intentions to change drinking behavior. The multigroup path model was run in the full sample and in the drinker-only sample. For all path models, we used self-efficacy and threat perception obtained at wave 2 (pre-feedback) to place these variables at the start of the model. Dummy variables were created to examine the effect of the two risk feedback groups (PHEN and PHEN+GENE) compared to the control group (CONTROL). An interaction term was also created from mean-centered versions of the threat perception and self-efficacy variables. We also included participants baseline heaviest week drinking quantity as a predictor of their change in heaviest week drinking quantity and controlled for the effects of age and sex on participants’ intentions to change their drinking behavior. We calculated the robust standard errors for all parameter estimates to account for the violations of normality assumptions in our data. We report the robust standard errors in the path model output and use the robust standard errors and an alpha level of < .050 to determine path significance. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 23 For indirect effects, we report the 95% CIs, with CIs that do not include 0 indicating a significant effect. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 24 Chapter 3: Results Descriptive Statistics Descriptive statistics for the heaviest week drinking quantity at waves 2-6 (split by ALDH2*2 status and feedback group) are presented in Table 2. Given the presence of some extreme values in the heaviest week drinking quantity data, we calculated the descriptive statistics in Table 2 after trimming the data at 5%. Furthermore, due to the high number of nondrinkers in the study (i.e. zero-inflated distribution), we present the proportion of the sample who were non-drinkers in the lefthand set of columns and the means and standard deviations for the drinker-only sample in the righthand set of columns. Table 3 shows the correlations, means, and standard deviations at each wave of the study for the outcome and predictor variables, including feedback group, threat perception, self-efficacy, intentions to change drinking behavior, and heaviest week drinking quantity. The Table 3 statistics were stratified by ALDH2*2 status and were calculated using data trimmed at 5%. Figure 2 depicts the longitudinal mean trajectories of the heaviest week drinking quantity variable over time for each ALDH2*2 status-feedback group condition. The data in Figure 2 were also trimmed at 5%. Effect of Predictors on Intentions to Change Drinking Behavior In our first set of multiple regression analyses, we examined the effects of our predictors of interest on participants’ intentions to change their drinking behavior. The parameter estimates for the model are shown in Table 4. Results for the total sample (first column of Table 4) showed that baseline heaviest week drinking quantity significantly predicted participants’ intentions to change their drinking behavior, such that the more participants drank during their heaviest drinking week at baseline, the more they intended to change their drinking behavior after viewing the feedback session (β = 0.06, p = .005). Feedback group also significantly predicted HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 25 intentions to change drinking behavior. Participants in PHEN (β = 0.67, p = .005) and PHEN+GENE (β = 0.63, p = .007) had significantly higher intentions to change their drinking compared to those who did not receive risk feedback (CONTROL). We did not find a significant interaction effect between ALDH2*2 status and feedback group, however, suggesting that the effect of risk feedback did not significantly differ between those with and without ALDH2*2. In line with this finding, splitting the sample by ALDH2*2 status (second and third columns of Table 4) revealed that receiving risk feedback was associated with significantly greater intentions to change drinking behavior among both ALDH2*2(+) participants (PHEN: β = 0.88, p = .001, or PHEN+GENE: β = 0.89, p = .001) and ALDH2*2(-) participants (PHEN: β = 0.72, p = .004, or PHEN+GENE β = 0.66, p = .006). We also found a significant interaction of Self-efficacy X Threat perception (β = 0.46, p = .006) so we ran the models stratifying by lower (score = 1 or 2; 67.5% of sample) and higher (score = 3 or 4, 32.5%) levels of threat perception (see Table 5). Results showed self-efficacy did not significantly predict intentions to change drinking behavior among those with low threat perception, but among participants with high threat perception, higher self-efficacy was associated with higher intentions to change drinking behavior (β = 0.60, p = .013; see first three columns of Table 5). When splitting the sample by ALDH2*2, the effects of threat perception and the Self-efficacy X Threat perception interaction were no longer statistically significant, although the effects of these variables were similar for those with and without ALDH2*2 with p = ~.100 (2nd and 3rd columns of Table 4). Repeating these analyses in the drinker-only sample largely revealed the same pattern of results, as shown in the last three columns of Table 4. Again, the interaction of Self-efficacy X Threat perception was significant (β = 0.70, p = .001), and in the drinker-only sample the HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 26 interaction remained significant after stratifying by ALDH2*2 status, both among participants with (β = 0.65, p = .028) and without (β = 0.61, p = .019) ALDH2*2. Stratifying the drinker-only sample by threat perception also mirrored what we found in the full sample (see last three columns of Table 5). Self-efficacy did not significantly predict intentions to change drinking behavior for drinkers with low threat perception, but higher self-efficacy significantly predicted greater intentions to change drinking behavior among drinkers with high threat perception (β = 0.79, p = .005). Effect of Intentions to Change Drinking Behavior on Change in Heaviest Week Drinking Quantity In our next set of analyses, we examined the extent to which risk feedback and participants’ intentions to change their drinking behavior were associated with a baseline to postfeedback change in their heaviest week drinking quantity at each wave following the feedback session. Parameter estimates for each wave and sample are presented in Table 6. Results were largely consistent across waves. Baseline heaviest week drinking quantity was the only consistent predictor of change in heaviest week drinking behavior across study waves. Those who reported higher baseline heaviest week drinking quantity had greater decreases in heaviest week drinking quantity from baseline to 1-, 4-, 7-, and 10-months postfeedback. Age significantly predicted increased heaviest week drinking quantity from baseline to 4-months post-feedback for those in the full sample (β = 0.28, p = .001), among ALDH2*2(-) participants (β = 0.49, p = .004), and among drinkers only (β = 0.34, p = .007), and decreased drinking from baseline to 7-months post-feedback among the full sample (β = -0.27, p = .039), among drinkers only (β = -0.61, p = .003), among ALDH2*2(-) drinkers (β = -0.90, p = .007), and among ALDH2*2(+) drinkers (β = -0.56, p = .014). Lastly, greater intentions to change HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 27 drinking behavior suggested an increase in heaviest week drinking quantity for ALDH2*2(-) participants from baseline to 10-months post- feedback (β = 0.33, p = .039) that was not significant after controlling for multiple comparisons using Hochberg’s method (Hochberg, 1988). Full Path Model In the final step of our analyses, we combined all our variables of interest into the full multigroup path model depicted in Figure 3 (see Table 7 for the parameter estimates and model fit statistics). Results from the full path model were consistent with our multiple regression findings above, showing a significant direct effect of feedback group (PHEN or PHEN+GENE compared to CONTROL) on intentions to change drinking behavior both among ALDH2*2(-) participants (PHEN: β = 0.70, p = .003, or PHEN+GENE: β = 0.71, p = .001) as well as ALDH2*2(+) participants (PHEN: β = 1.01, p < .001, or PHEN+GENE: β = 0.91, p < .001). Higher baseline heaviest week drinking quantity also predicted greater decreases in heaviest week drinking from baseline to 1-month post-feedback among both ALDH2*2(-) participants (β = -0.42, p < .001) and ALDH2*2(+) participants (β = -0.37, p = .002). No other individual paths in the model reached significance. Lastly, we found an indirect effect of receiving risk feedback on change in heaviest week drinking quantity via intentions to change drinking behavior that approached significance among ALDH2*2(-) participants (PHEN: β = 0.23, 95% CI: [0.00, 0.60], or PHEN+GENE: β = 0.24, 95% CI: [0.00, 0.61]). We did not find that the same indirect path approached significance among ALDH2*2(+) participants (PHEN: β = 0.14, 95% CI: [-0.15, 0.48]; PHEN+GENE: β = 0.13, 95% CI: [-0.13, 0.45]). Figure 4 depicts the full path model rerun in the drinker-only sample (see Table 8 for the parameter estimates and model fit statistics). Results for the drinker-only sample were consistent HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 28 with those for the full sample, showing that receiving the risk feedback was associated with greater intentions to change drinking compared to receiving the CONTROL, both among participants without ALDH2*2 (PHEN: β = 0.90, p = .001, or PHEN+GENE: β = 0.79, p = .001) and among those with ALDH2*2 (PHEN: β = 1.09, p < .001, or PHEN+GENE: β = 1.25, p < .001). Higher levels of baseline heaviest week drinking predicted a decrease in heaviest week drinking behavior from baseline to 1-month post- feedback among ALDH2*2(-) participants (β = -0.46, p < .001) and ALDH2*2(+) participants (β = -0.43, p = .001). In the drinker-only sample, being male was associated with lower intentions to change drinking (β = -0.41, p = .046). We again found the trend of an indirect effect of receiving risk feedback (compared to receiving CONTROL feedback) predicting greater intentions to change drinking behavior, which then led to an increase in heaviest week drinking quantity from baseline to 1-month post-feedback among ALDH2*2(-) participants (PHEN: β = 0.36, 95% CI: [-0.03, 0.91], or PHEN+GENE: β = 0.32, 95% CI: [-0.03, 0.82]), although this trend did not reach statistical significance. As in the full sample, we did not find evidence for the same indirect effect among ALDH2*2(+) drinkers (PHEN: β = 0.13, 95% CI: [-0.27, 0.57]; PHEN+GENE: β = 0.15, 95% CI: [-0.30, 0.65]). HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 29 Chapter 4: Discussion The purpose of the current study was to evaluate the effects of a brief online feedback intervention on intentions to change drinking behavior and actual change in heaviest week drinking quantity over the course of the study. The feedback intervention informed participants that drinking alcohol increases risk for certain cancers, and that possessing an ALDH2*2 allele further increases cancer risk among those who drink alcohol. Our first aim was to evaluate the extent to which at-risk individuals (those who have an ALDH2*2 allele and drink alcohol) reported greater intentions to change their drinking behavior after learning about their elevated risk. Our second aim was to examine whether two proposed mechanisms of behavior change, self-efficacy and threat perception, played a role in participants’ intentions to change their drinking behavior. Furthermore, we explored whether receiving risk feedback mediated the extent to which participants’ baseline self-efficacy and threat perception related to their intentions to change their drinking behavior. Our last aim was to examine whether participants’ intentions to change their drinking behavior was associated with actual changes in their heaviest week drinking quantity over the course of the study. We found support for our hypothesis that among participants with ALDH2*2, those who received information about the link between alcohol, flushing, ALDH2*2, and cancer risk had greater intentions to change their drinking behavior compared to those who did not receive risk feedback. This effect was found for both types of risk feedback, meaning participants were motivated to change their drinking behavior whether they received information about the risk associated with flushing alone (phenotype feedback only) or in combination with their own ALDH2 genotype (phenotype and genotype feedback). This finding suggests that risk information about the flushing phenotype motivates participants to change their drinking, and HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 30 that additional genetic feedback may not greatly enhance this effect. Furthermore, the risk feedback was associated with greater intentions to change drinking behavior regardless of whether participants had an ALDH2*2 allele. This finding is consistent with Hendershot and colleagues’ (2010) finding that receiving the risk feedback increased participants’ intentions to change their drinking among those without ALDH2*2, even when they did not find this effect among those with ALDH2*2. It is possible that in our study the general feedback that alcohol consumption increases risk for five types of cancer was enough to motivate participants to reduce their drinking, even among those who were not at increased genetic risk. Future research could disentangle the effect of the personalized feedback about flushing and ALDH2*2 from the more general feedback that alcohol use increases cancer risk by including a feedback condition that only provides information on the cancer risk associated with alcohol use. Additionally, a proportion of people who do not have an ALDH2*2 allele still flush from drinking alcohol, likely due to other alcohol metabolizing gene variants (Luczak et al., 2011). People who do not have the ALDH2*2 allele but who still flush from alcohol might come to different conclusions about their risk depending on whether they received the phenotype feedback based on flushing, or the phenotype plus genotype feedback based on flushing and their own ALDH2 genotype. A more nuanced pattern of results emerges when splitting the sample by flushing status as well as ALDH2*2 status (see Luczak et al., 2024). We also found evidence to support our second hypothesis that threat perception, selfefficacy, and their interaction (measured at 1-month post-intervention) are associated with greater intentions to change drinking behavior. Among participants with higher threat perception, higher self-efficacy was associated with greater intentions to change drinking behavior, whereas among those with lower threat perception, self-efficacy was not related to intentions to change HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 31 drinking behavior. This finding aligns with prior literature and suggests that if threat perception is too low, people will not be motivated to change their behavior, rendering their degree of selfefficacy, or belief that they can change their behavior, less relevant (Rogers, 1975). Our results are also consistent with findings showing that in the context of health messaging campaigns, low threat messages have the least impact on attitudes, intentions, and behaviors regardless of efficacy content, while high threat/high efficacy messages have the greatest impact on behavior (Witte & Allen, 2000). In our exploratory analyses, we did not find that participants’ baseline levels of threat perception and self-efficacy indirectly impacted participants’ intentions to change their drinking behavior via receiving risk feedback. Taken together, our findings suggest that it is the interaction between self-efficacy and threat perception post-feedback (but not pre-feedback) that relates to intentions to change drinking behavior. In Hendershot et al. (2010), those who received risk feedback had greater increases in pre- to post-intervention threat perception compared to those who did not receive risk feedback, both among participants with ALDH2*2 (i.e., cancer risk) and without ALDH2*2 (i.e., alcohol use disorder risk). Our future studies will obtain measures of self-efficacy and threat perception immediately after the risk feedback to allow us to examine the degree to which risk feedback affects participants’ levels of self-efficacy and threat perception over time, as well as the degree to which these variables account for the relationship between receiving the risk information and intending to change one’s drinking behavior. Our findings did not support our final hypothesis--intentions to change drinking behavior did not significantly predict decreased heaviest week drinking quantity from baseline to 1-, 4-, 7- , or 10-months post-feedback. Of the two previous studies that have examined the effects of brief genetic feedback interventions for ALDH2*2 and alcohol use, one found significant intervention HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 32 effects on drinking frequency and typical drinking quantity measured 1-month post-feedback (Hendershot et al., 2010) and the other did not find significant intervention effects on drinking frequency 18-months post-feedback (Komiya et al., 2006). Similar to Komiya and colleagues (2006), our study did not find that the risk feedback was linked to significant changes in drinking behavior over the following year, but did find non-significant trends suggesting that drinking frequency decreased in those with ALDH2*2 who were given risk feedback and increased in all other groups (those without ALDH2*2 and those with ALDH2*2 who did not receive risk feedback). We also note that the methodology differed across these studies, for example our study sample differed substantially from Komiya and colleagues’ sample in that our participants were young-adult students whose drinking trajectories are expected to increase over the course of their time in college, while their participants were Japanese employees at a manufacturing company and were largely in their 30s and 40s (2006). Additionally, we only examined baseline to post-feedback change in heaviest week drinking quantity as a drinking outcome variable, whereas the two prior studies examined different alcohol outcome measures (e.g., alcohol use frequency, maximum drinks consumed, typical number of drinks on weekend nights). It is possible that additional analyses more fully capturing drinking behavior will reveal additional intervention effects. It is also possible that including booster sessions reminding participants of the feedback material would increase study effects on alcohol consumption levels. We also found a non-significant indirect effect in our path models, in which participants without ALDH2*2 who received risk feedback had higher intentions to change drinking behavior compared to those who did not receive risk feedback, which was then associated with increased drinking from baseline to 1-month post-feedback. Participants without ALDH2*2 may have felt they did not need to decrease their drinking after learning they have lower risk for alcohol- HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 33 related cancers compared to their peers who flush and/or possess an ALDH2*2 allele. In general, however, college student drinking tends to increase on average during the college years, including in east Asian college students with and without ALDH2*2 (Luczak et al., 2014). Our finding may therefore reflect a normative increase in drinking behavior among participants without ALDH2*2 who learned they are not at increased genetic risk. It should also be noted that participants’ heaviest week drinking quantity fluctuated greatly over the course of the school year (see Figure 2), for example, decreasing over the summer when students were largely offcampus at wave 4 and increasing as they returned to campus in the fall at wave 5. Future studies could better parse the effects of a feedback intervention from natural patterns of college drinking by continuing to follow participants over longer durations of time and accounting for typical month-to-month fluctuations in drinking patterns (e.g., Del Boca et al., 2004; Luczak et al., 2014). Nevertheless, the possibility of an unintended consequence of the risk feedback--that participants might drink more after being notified they are not at increased risk for alcoholrelated cancers, even if they might be at increased risk for alcohol dependence--warrants further attention in futures studies. The content of the risk feedback provided in this study and in the two prior ALDH2*2 studies (Hendershot et al., 2010; Komiya et al., 2006) focus on distal alcohol-related health consequences (cancer risk, alcohol use disorder), which may not feel particularly urgent to college students. This contrasts with many other brief interventions for alcohol use in college student samples, which often provide normative feedback about peer drinking levels (Carey et al., 2009) that students may be more likely to act on in a shorter time frame. Participants in our study may still be experimenting with alcohol and may view their current drinking behavior as specific to the college environment rather than as a lasting habit that could result in long-term HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 34 health consequences. Future research could therefore explore the effects of interventions about alcohol-related cancer risk over longer follow-up periods or in older samples in which participants’ drinking has begun to plateau or decrease and when long-term health consequences like cancer risk may feel more salient. Finally, some participants also may not have reduced their drinking after viewing our feedback sessions because they did not drink heavily enough to be concerned about the health consequences. In our study, 88% of participants reported lifetime drinking, and 40% of participants reported at least one past 30-day binge drinking episode (4+ standard drinks for females and 5+ standard drinks for males in a single drinking episode) just prior to the feedback session. Hendershot and colleagues reported a similar proportion of lifetime drinkers (90%) and binge drinkers (39%) in their sample (2010). Despite these rates of binge drinking, however, it is possible that participants did not view their drinking habits as concerning given the information we shared with them. On the one hand, our feedback sessions taught participants that cancer risk increases with the amount of alcohol consumed, and that the surest way to reduce risk is to reduce or eliminate alcohol use. On the other hand, the individual studies we highlighted used relatively high cutoffs for what was considered heavy drinking that would increase cancer risk. Boccia and colleagues (2009), for example, reported that heavy drinkers with an ALDH2*2 allele were 3.57 times more likely to develop head and neck cancers than heavy drinkers with no variant alleles, but the cutoff for heavy drinking in this meta-analysis was 59 grams of alcohol per day, which is 4.21 standard drinks in the U.S. Few participants in our study came close to this level of drinking, and therefore our participants may not have felt motivated to change their behavior as a result of this information. Our series of analyses did examine drinkers-only and HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 35 included baseline drinking in the models, but future research could examine more nuanced relationships among lighter and heavier drinkers. In addition to the limitations already mentioned, our study had relatively small subgroups within our larger sample of 360 participants, with only 55 participants receiving feedback that they had an ALDH2*2 allele. Results from this pilot study should therefore be considered first steps in understanding the relationships among these feedback sessions, predictors, and outcomes. Additionally, the timeline of our study was impacted by the COVID-19 pandemic. Our first wave of data collection began as students were returning to campus after the pandemic moved classes to online learning. It is possible that future studies conducted under more normal social settings will reveal different findings. Further research is needed to increase our understanding of the effects of this type of risk feedback on drinking behavior and to elucidate the relationship between intentions to change drinking behavior and actual drinking behavior. In particular, longer-term longitudinal studies could extend this work by disentangling the normative fluctuations and progression of drinking during the college years from the effects of the risk feedback, as well as to assess for dormant intervention effects that may manifest after participants leave college. Future replications of this study could also be conducted in samples of heavy drinkers who are at higher risk for alcoholrelated cancers and in samples of older participants who may be more motivated to protect themselves against distal health risks. Overall, our study demonstrates that informing people about the cancer risk associated with drinking alcohol, flushing, and having an ALDH2*2 allele was associated with greater intentions to change drinking behavior. Furthermore, our findings suggest that informing participants of their own personalized ALDH2*2 genotype did not meaningfully enhance the HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 36 intervention’s effect on participants intentions to change their drinking behavior over and above informing them of the general risks associated with drinking, flushing, and ALDH2*2. This novel finding indicates potential for delivering the intervention at a wide scale without the added cost of genotyping each individual intervention recipient. Our findings also suggest that two mechanisms of behavior change, self-efficacy and threat perception, play a role in people’s intentions to change their drinking behavior, which sheds light on why some participants may be motivated to act on this health risk information while others are not. We did not, however, find that higher intentions to change drinking behavior led to decreased alcohol consumption during participants’ heaviest drinking week over the course of the study. Additional research is therefore needed to elucidate how intentions to change drinking behavior can be translated into actual changes in behavior over time. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 37 References Ajzen, I. (1985). From Intentions to Actions: A theory of planned behavior. In J. Kuhl & J. Beckmann (Eds.), Action Control: From Cognition to Behavior (pp. 11–39). Springer. https://doi.org/10.1007/978-3-642-69746-3_2 Ajzen, I. (1987). Attitudes, traits, and actions: Dispositional prediction of behavior in personality and social psychology. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 20, pp. 1–63). Academic Press. https://doi.org/10.1016/S0065- 2601(08)60411-6 Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211. https://doi.org/10.1016/0749-5978(91)90020-T Armitage, C. J., Armitage, C. J., Conner, M., Loach, J., & Willetts, D. (1999). Different perceptions of control: Applying an extended theory of planned behavior to legal and illegal drug use. Basic and Applied Social Psychology, 21(4), 301–316. https://doi.org/10.1207/S15324834BASP2104_4 Armitage, C. J., & Conner, M. (1999). Distinguishing perceptions of control from self-efficacy: Predicting consumption of a low-fat diet using the theory of planned behavior. Journal of Applied Social Psychology, 29(1), 72–90. https://doi.org/10.1111/j.1559- 1816.1999.tb01375.x Audrain, J., Boyd, N. R., Roth, J., Main, D., Caporaso, N. E., & Lerman, C. (1997). Genetic susceptibility testing in smoking-cessation treatment: One-year outcomes of a randomized trial. Addictive Behaviors, 22(6), 741–751. https://doi.org/10.1016/S0306- 4603(97)00060-9 Baan, R., Straif, K., Grosse, Y., Secretan, B., El Ghissassi, F., Bouvard, V., Altieri, A., & Cogliano, V. (2007). Carcinogenicity of alcoholic beverages. The Lancet Oncology, 8(4), 292–293. https://doi.org/10.1016/S1470-2045(07)70099-2 Bagnardi, V., Rota, M., Botteri, E., Tramacere, I., Islami, F., Fedirko, V., Scotti, L., Jenab, M., Turati, F., Pasquali, E., Pelucchi, C., Galeone, C., Bellocco, R., Negri, E., Corrao, G., Boffetta, P., & La Vecchia, C. (2015). Alcohol consumption and site-specific cancer risk: A comprehensive dose–response meta-analysis. British Journal of Cancer, 112(3), 580– 593. https://doi.org/10.1038/bjc.2014.579 Bates, D., Maechler, M., Bolker, B., Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1-48. doi:10.18637/jss.v067.i01. Boker, S. M., Neale, M. C., Maes, H. H., Spiegel, M., Brick, T. R., Estabrook, R., Bates, T. C., Gore, R. J., Hunter, M. D., Pritikin, J. N., Zahery, M., Kirkpatrick, R. M. (2023). OpenMx: Extended structural equation modelling. R package version 2.21.8, <https://CRAN.R-project.org/package=OpenMx>. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 38 Boccia, S., Hashibe, M., Gallì, P., De Feo, E., Asakage, T., Hashimoto, T., Hiraki, A., Katoh, T., Nomura, T., Yokoyama, A., van Duijn, C. M., Ricciardi, G., & Boffetta, P. (2009). Aldehyde dehydrogenase 2 and head and neck cancer: A meta-analysis implementing a mendelian randomization approach. Cancer Epidemiology, Biomarkers & Prevention, 18(1), 248–254. https://doi.org/10.1158/1055-9965.EPI-08-0462 Brooks, P. J., Enoch, M.-A., Goldman, D., Li, T.-K., & Yokoyama, A. (2009). The alcohol flushing response: An unrecognized risk factor for esophageal cancer from alcohol consumption. PLOS Medicine, 6(3), e1000050. https://doi.org/10.1371/journal.pmed.1000050 Carey, K. B., Scott-Sheldon, L. A. J., Elliott, J. C., Bolles, J. R., & Carey, M. P. (2009). Computer-delivered interventions to reduce college student drinking: A meta-analysis. Addiction, 104(11), 1807–1819. https://doi.org/10.1111/j.1360-0443.2009.02691.x Chang, J. S., Hsiao, J.-R., & Chen, C.-H. (2017). ALDH2 polymorphism and alcohol-related cancers in Asians: A public health perspective. Journal of Biomedical Science, 24(1), 19. https://doi.org/10.1186/s12929-017-0327-y Collins, R., Parks, G., & Marlatt, G. (1985). Social determinants of alcohol consumption: The effects of social interaction and model status on the self-administration of alcohol. Journal of Consulting and Clinical Psychology, 53, 189–200. https://doi.org/10.1037/0022-006X.53.2.189 Collins, S. E., Carey, K. B., & Sliwinski, M. J. (2002). Mailed personalized normative feedback as a brief intervention for at-risk college drinkers. Journal of Studies on Alcohol, 63(5), 559–567. https://doi.org/10.15288/jsa.2002.63.559 Conner, M., Warren, R., Close, S., & Sparks, P. (1999). Alcohol consumption and the theory of planned behavior: An examination of the cognitive mediation of past behavior. Journal of Applied Social Psychology, 29(8), 1676–1704. https://doi.org/10.1111/j.1559- 1816.1999.tb02046.x Del Boca, F. K., Darkes, J., Greenbaum, P. E., & Goldman, M. S. (2004). Up close and personal: Temporal variability in the drinking of individual college students during their first year. Journal of consulting and clinical psychology, 72(2), 155. Doumas, D. M., McKinley, L. L., & Book, P. (2009). Evaluation of two web-based alcohol interventions for mandated college students. Journal of Substance Abuse Treatment, 36(1), 65–74. https://doi.org/10.1016/j.jsat.2008.05.009 Eng, M. Y., Luczak, S. E., & Wall, T. L. (2007). ALDH2, ADH1B, and ADH1C genotypes in Asians: A literature review. Alcohol Research & Health, 30(1), 22–27. HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 39 Goedde, H. W., Agarwal, D. P., Fritze, G., Meier-Tackmann, D., Singh, S., Beckmann, G., Bhatia, K., Chen, L. Z., Fang, B., Lisker, R., Paik, Y. K., Rothhammer, F., Saha, N., Segal, B., Srivastava, L. M., & Czeizel, A. (1992). Distribution of ADH2 and ALDH2 genotypes in different populations. Human Genetics, 88(3). https://doi.org/10.1007/BF00197271 Guttmacher, A. E., & Collins, F. S. (2005). Realizing the promise of genomics in biomedical research. JAMA, 294(11), 1399–1402. https://doi.org/10.1001/jama.294.11.1399 Hendershot, C. S., Otto, J. M., Collins, S. E., Liang, T., & Wall, T. L. (2010). Evaluation of a brief web-based genetic feedback intervention for reducing alcohol-related health risks associated with ALDH2. Annals of Behavioral Medicine, 40(1), 77–88. https://doi.org/10.1007/s12160-010-9207-3 Henrikson, N. B., Bowen, D., & Burke, W. (2009). Does genomic risk information motivate people to change their behavior? Genome Medicine, 1(3), 37. https://doi.org/10.1186/gm37 Hietaranta-Luoma, H. L., Tahvonen, R., Iso-Touru, T., Puolijoki, H., Hopia, A. An intervention study of individual, apoE genotype-based dietary and physical-activity advice: Impact on health behavior. Journal of nutrigenetics and nutrigenomics. 2014;7(3):161-174. doi:10.1159/000371743 Hochberg, Y. (1988). A sharper Bonferroni procedure for multiple tests of significance. Biometrika, 75(4), 800-802. Hollands, G. J., French, D. P., Griffin, S. J., Prevost, A. T., Sutton, S., King, S., & Marteau, T. M. (2016). The impact of communicating genetic risks of disease on risk-reducing health behaviour: Systematic review with meta-analysis. BMJ, i1102. https://doi.org/10.1136/bmj.i1102 Horne, J., Madill, J., O’Connor, C., Shelley, J., & Gilliland, J. (2018). A systematic review of genetic testing and lifestyle behaviour change: Are we using high-quality genetic interventions and considering behaviour change theory?. Lifestyle Genomics, 11(1), 49- 63. Huh, D., Mun, E.-Y., Larimer, M. E., White, H. R., Ray, A. E., Rhew, I. C., Kim, S.-Y., Jiao, Y., & Atkins, D. C. (2015). Brief motivational interventions for college student drinking may not be as powerful as we think: An individual participant-level data meta-analysis. Alcohol: Clinical and Experimental Research, 39(5), 919–931. https://doi.org/10.1111/acer.12714 Jorgensen, T. D., Pornprasertmanit, S., Schoemann, A. M., & Rosseel, Y. (2022). semTools: Useful tools for structural equation modeling. R package version 0.5-6. Retrieved from https://CRAN.R-project.org/package=semTools HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 40 Knight, J. R., Wechsler, H., Kou, M., Seibering, M., Weitzman, E. R., & Schuckit, M. A. (2002). Alcohol abuse and dependence among U.S. college students. Journal of Studies on Alcohol, 63(3), 263–271. Komiya, Y., Nakao, H., Kuroda, Y., Arizono, K., Nakahara, A., & Katoh, T. (2006). Application of aldehyde dehydrogenase 2 (ALDH2) genetic diagnosis in support of decreasing alcohol intake. Journal of Occupational Health, 48(3), 161–165. https://doi.org/10.1539/joh.48.161 Larimer, M. E., Cronce, J. M., Lee, C. M., & Kilmer, J. R. (2004). Brief intervention in college settings. Alcohol Research & Health, 28(2), 94–104. Larimer, M. E., Lee, C. M., Kilmer, J. R., Fabiano, P. M., Stark, C. B., Geisner, I. M., Mallett, K. A., Lostutter, T. W., Cronce, J. M., Feeney, M., & Neighbors, C. (2007). Personalized mailed feedback for college drinking prevention: A randomized clinical trial. Journal of Consulting and Clinical Psychology, 75, 285–293. https://doi.org/10.1037/0022- 006X.75.2.285 Lerman, C., Gold, K., Audrain, J., Lin, T. H., Boyd, N. R., Wilfond, B., & Louben, G. (1997). Incorporating biomarkers of exposure and genetic susceptibility into smoking cessation treatment: effects on smoking-related cognitions, emotions, and behavior change. Health Psychology, 16(1), 87. Li, T. K. (2000). Pharmacogenetics of responses to alcohol and genes that influence alcohol drinking. Journal of Studies on Alcohol, 61(1), 5–12. https://doi.org/10.15288/jsa.2000.61.5 Luczak, S. E., Pandika, D., Shea, S. H., Eng, M. Y., Liang, T., & Wall, T. L. (2011). ALDH2 and ADH1B interactions in retrospective reports of low-dose reactions and initial sensitivity to alcohol in Asian American college students. Alcohol: Clinical and Experimental Research, 35(7), 1238–1245. https://doi.org/10.1111/j.1530- 0277.2011.01458.x Luczak, S. E., Saldich, E. B., Monterosso, J. R., Lai, M. H. C., Culver, J. O., Witkiewitz, K., & Hendershot C. S. (2024). Ten-month outcomes of brief online interventions that provide personalized feedback on alcohol metabolism and disease risk to Asian American college students. Alcoholism: Clinical and Experimental Research, 48, (Suppl. 1), A. Luczak, S. E., Yarnell, L. M., Prescott, C. A., Myers, M. G., Liang, T., & Wall, T. L. (2014). Effects of ALDH2*2 on alcohol problem trajectories of Asian American college students. Journal of abnormal psychology, 123(1), 130. Marlatt, G. A., Baer, J. S., Kivlahan, D. R., Dimeff, L. A., Larimer, M. E., Quigley, L. A., Somers, J. M., & Williams, E. (1998). Screening and brief intervention for high-risk college student drinkers: Results from a 2-year follow-up assessment. Journal of Consulting and Clinical Psychology, 66, 604–615. https://doi.org/10.1037/0022- 006X.66.4.604 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 41 McBride, C. M., Bepler, G., Lipkus, I. M., Lyna, P., Samsa, G., Albright, J., Datta, S., & Rimer, B. K. (2002). Incorporating genetic susceptibility feedback into a smoking cessation program for African-American smokers with low income. Cancer Epidemiology, Biomarkers & Prevention, 11(6), 521–528. McCarthy, M. I., & Hirschhorn, J. N. (2008). Genome-wide association studies: Potential next steps on a genetic journey. Human Molecular Genetics, 17(R2), R156–R165. https://doi.org/10.1093/hmg/ddn289 McCaul, K. D., Sandgren, A. K., O’Neill, H. K., & Hinsz, V. B. (1993). The value of the theory of planned behavior, perceived control, and self-efficacy expectations for predicting health-protective behaviors. Basic and Applied Social Psychology, 14(2), 231–252. https://doi.org/10.1207/s15324834basp1402_7 Mcmillan, B., & Conner, M. (2003). Applying an extended version of the theory of planned behavior to illicit drug use among students. Journal of Applied Social Psychology, 33(8), 1662–1683. https://doi.org/10.1111/j.1559-1816.2003.tb01968.x Milton, A. C., & Mullan, B. A. (2012). An application of the theory of planned behavior—A randomized controlled food safety pilot intervention for young adults. Health Psychology, 31, 250–259. https://doi.org/10.1037/a0025852 Naimi, B., Hamm, N. A., Groen, T. A., Skidmore, A. K., Toxopeus, A. G. (2014). Ecography: Where is positional uncertainty a problem for species distribution modelling. R package version 2.1.7, 191-203. doi:10.1111/j.1600-0587.2013.00205.x. <https://doi.org/10.1111/j.1600-0587.2013.00205.x>. Neighbors, C., Lewis, M. A., Atkins, D. C., Jensen, M. M., Walter, T., Fossos, N., Lee, C. M., & Larimer, M. E. (2010). Efficacy of web-based personalized normative feedback: A twoyear randomized controlled trial. Journal of Consulting and Clinical Psychology, 78, 898–911. https://doi.org/10.1037/a0020766 Newman, I. M., Shell, D. F., Huang, Z., & Qian, L. (2015). The educational potential of alcoholrelated flushing among Chinese young people. Drugs: Education, Prevention and Policy, 22(1), 52–59. https://doi.org/10.3109/09687637.2014.979764 R Core Team. (2022). R: The R Project for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.r-project.org/ Revelle, W. (2023). psych: Procedures for Psychological, Psychometric, and Personality Research. Northwestern University, Evanston, Illinois. R package version 2.3.6, <https://CRAN.R-project.org/package=psych>. Rimal, R. N., & Real, K. (2003). Perceived risk and efficacy beliefs as motivators of change. Human Communication Research, 29(3), 370–399. https://doi.org/10.1111/j.1468- 2958.2003.tb00844.x HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 42 Rochefort, C., Baldwin, A. S., Tiro, J., & Bowen, M. E. (2020). Evaluating the validity of the risk perception survey for developing diabetes scale in a safety-net clinic population of English and Spanish speakers. The Diabetes Educator, 46(1), 73–82. https://doi.org/10.1177/0145721719889068 Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change. The Journal of Psychology, 91(1), 93–114. https://doi.org/10.1080/00223980.1975.9915803 Sanderson, S. C., Humphries, S. E., Hubbart, C., Hughes, E., Jarvis, M. J., & Wardle, J. (2008). Psychological and behavioural impact of genetic testing smokers for lung cancer risk: A phase II exploratory trial. Journal of Health Psychology, 13(4), 481–494. https://doi.org/10.1177/1359105308088519 Schnoll, R. A., Wang, H., Miller, S. M., Babb, J. S., Cornfeld, M. J., Tofani, S. H., HenniganPeel, T., Balshem, A., Slater, E., Ross, E., Boyd, C. S., & Engstrom, P. F. (2005). Change in worksite smoking behavior following cancer risk feedback: A pilot study. American Journal of Health Behavior, 29(3), 215–227. https://doi.org/10.5993/AJHB.29.3.3 Schwarzer, R. (2008). Modeling health behavior change: How to predict and modify the adoption and maintenance of health behaviors. Applied Psychology, 57(1), 1–29. https://doi.org/10.1111/j.1464-0597.2007.00325.x Schwarzer, R., & Fuchs, R. (1995). Changing risk behaviors and adopting health behaviors: The role of self-efficacy beliefs. In A. Bandura (Ed.), Self-Efficacy in Changing Societies (1st ed., pp. 259–288). Cambridge University Press. https://doi.org/10.1017/CBO9780511527692.011 Silarova, B., Sharp, S., Usher-Smith, J. A., Lucas, J., Payne, R. A., Shefer, G., Moore, C., Girling, C., Lawrence, K., Tolkien, Z., Walker, M., Butterworth, A., Angelantonio, E. D., Danesh, J., & Griffin, S. J. (2019). Effect of communicating phenotypic and genetic risk of coronary heart disease alongside web-based lifestyle advice: The INFORM Randomised Controlled Trial. Heart, 105(13), 982–989. https://doi.org/10.1136/heartjnl2018-314211 Sobell, L. C., & Sobell, M. C. (1992). Time-Line Follow-Back: A technique for assessing selfreported alcohol consumption. In R. Z. Litten & J. P. Allens (Eds.), Measuring alcohol consumption (pp. 73–98). Totowa, NJ: Humana Press. Sparks, J. A., Iversen, M. D., Yu, Z., Triedman, N. A., Prado, M. G., Miller Kroouze, R., Kalia, S. S., Atkinson, M. L., Mody, E. A., Helfgott, S. M., Todd, D. J., Dellaripa, P. F., Bermas, B. L., Costenbader, K. H., Deane, K. D., Lu, B., Green, R. C., & Karlson, E. W. (2018). Disclosure of personalized rheumatoid arthritis risk using genetics, biomarkers, and lifestyle factors to motivate health behavior improvements: A randomized controlled trial. Arthritis Care & Research, 70(6), 823–833. https://doi.org/10.1002/acr.23411 Venables, W. N. & Ripley, B. D. (2002) Modern Applied Statistics with S. Fourth Edition. Springer, New York. ISBN 0-387-95457-0 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 43 Walker, E. A., Mertz, C. K., Kalten, M. R., & Flynn, J. (2003). Risk perception for developing diabetes: Comparative risk judgments of physicians. Diabetes Care, 26(9), 2543–2548. https://doi.org/10.2337/diacare.26.9.2543 White, A., Kavanagh, D., Stallman, H., Klein, B., Kay-Lambkin, F., Proudfoot, J., Drennan, J., Connor, J., Baker, A., Hines, E., & Young, R. (2010). Online alcohol interventions: A systematic review. Journal of Medical Internet Research, 12(5), e1479. https://doi.org/10.2196/jmir.1479 Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. Wickham H., Hester J., Chang W., Bryan J. (2022). devtools: Tools to Make Developing R Packages Easier. R package version 2.4.5, <https://CRAN.Rproject.org/package=devtools>. Wickham, H., Vaughan, D., Girlich, M. (2023). tidyr: Tidy Messy Data. R package version 1.3.0, <https://CRAN.R-project.org/package=tidyr>. Witte, K. (1992). Putting the fear back into fear appeals: The extended parallel process model. Communication Monographs, 59(4), 329. https://doi.org/10.1080/03637759209376276 Witte, K., & Allen, M. (2000). A meta-analysis of fear appeals: Implications for effective public health campaigns. Health Education & Behavior, 27(5), 591–615. https://doi.org/10.1177/109019810002700506 World Health Organization. (2018). Global status report on alcohol and health 2018. https://www.who.int/publications-detail-redirect/9789241565639 Wright, A. J., French, D. P., Weinman, J., & Marteau, T. M. (2006). Can genetic risk information enhance motivation for smoking cessation? An analogue study. Health Psychology, 25(6), 740–752. https://doi.org/10.1037/0278-6133.25.6.740 Zhao, T., Wang, C., Shen, L., Gu, D., Xu, Z., Zhang, X., Xu, Y., & Chen, J. (2015). Clinical significance of ALDH2 rs671 polymorphism in esophageal cancer: Evidence from 31 case-control studies. OncoTargets and Therapy, 8, 649–659. https://doi.org/10.2147/OTT.S76526 HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 44 Table 1 Study Procedure Wave 1 Aug - Oct Wave 2 February Wave 3 March Wave 4 June Wave 5 September Wave 6 December Initial Session Feedback Session 1-month follow-up 4-month follow-up 7-month follow-up 10-month follow-up Description 1 hour in-person: -Informed consent -DNA sample -online survey completed in lab 1 hour online: -10-minute online repeated survey -Feedback Session -Post-feedback survey 10-minute online repeated survey 10-minute online repeated survey 10-minute online repeated survey 20-minute online repeated survey Participants: Assigned n = 360 n = 360 n = 324 n = 324 n = 324 n = 324 Completed n = 360 n = 324 n = 310 n = 287 n = 295 n = 283 (% Female) (58%) (58%) (59%) (59%) (60%) (59%) Measures: Alcohol use X X (pre- feedback) X X X X Alcohol use Intentions to change X (post- feedback) Self-efficacy X X (pre- feedback) X X X X Threat perception X X (pre- feedback) X X X X HEALTH BEHAVIOR: SELF-EFFICACY AND PERCEIVED THREAT 45 Table 2 Means and Standard Deviations of Heaviest Week Drinking Quantity by Wave and Feedback Group with Data Trimmed at 5% % = proportion of non-drinkers for ALDH2*2 status by feedback group cell Feedback Group Non-Drinkers Drinkers ALDH2*2(-) n (%) ALDH2*2(+) n (%) ALDH2*2(-) Mean (SD), n ALDH2*2(+) Mean (SD), n Wave 2 CONTROL 4 (9%) 11 (25%) 2.35 (2.128), 43 1.65 (2.282), 33 PHEN 16 (33%) 9 (18%) 2.91 (3.202), 32 1.85 (2.008), 40 PHEN+GENE 12 (22%) 10 (22%) 2.61 (2.449), 42 2.03 (2.044), 35 Total 62 (22%) 2.24 (2.337), 225 Feedback Session (directly after wave 2) Wave 3 CONTROL 3 (6%) 8 (21%) 2.71 (2.471), 45 1.71 (2.076), 30 PHEN 14 (33%) 8 (17%) 1.96 (2.629), 29 2.25 (2.575), 40 PHEN+GENE 10 (20%) 10 (23%) 2.54 (2.457), 39 1.54 (1.713), 33 Total 53 (20%) 2.17 (2.371), 216 Wave 4 CONTROL 3 (7%) 8 (23%) 2.19 (2.413), 39 1.07 (1.266), 27 PHEN 15 (38%) 7 (16%) 2.46 (2.453), 24 0.78 (0.995), 37 PHEN+GENE 8 (20%) 9 (24%) 1.61 (1.948), 32 1.27 (1.867), 29 Total 50 (21%) 1.54 (1.961), 188 Wave 5 CONTROL 4 (10%) 6 (18%) 4.00 (3.397), 36 2.71 (2.291), 28 PHEN 13 (28%) 9 (20%) 3.36 (3.380), 33 2.32 (2.849), 37 PHEN+GENE 10 (21%) 10 (26%) 3.88 (3.124), 38 2.61 (2.779), 28 Total 52 (21%) 3.19 (3.057), 200 Wave 6 CONTROL 3 (8%) 7 (20%) 2.49 (3.081), 35 2.11 (2.644), 28 PHEN 13 (33%) 8 (18%) 2.67 (2.675), 27 1.82 (2.112), 36 PHEN+GENE 10 (22%) 9 (23%) 3.40 (3.240), 36 2.07 (2.392), 30 Total 50 (21%) 2.44 (2.748), 192 Table 3 Correlations for Feedback Group, Intentions to Change Drinking Behavior, Heaviest Week Drinking Quantity, Threat Perception, and Self-Efficacy at Waves 2 – 6 Split on the Diagonal by ALDH2*2 Status with Data Trimmed at 5% Feedback Intent HWD2 HWD3 HWD4 HWD5 HWD6 SE2 SE3 SE4 SE5 SE6 TP2 TP3 TP4 TP5 TP6 Mean - 1.27 1.38 1.48 0.78 1.91 1.52 1.95 1.91 1.91 1.81 1.94 1.17 1.28 1.22 1.18 1.20 SD - 1.34 1.98 2.16 1.29 2.54 2.22 0.64 0.64 0.60 0.61 0.62 0.66 0.64 0.69 0.71 0.58 Feedback 1 0.28 0.08 -0.05 0.06 -0.05 0.01 -0.08 -0.04 -0.08 0.02 -0.01 -0.08 0.01 -0.03 0.09 -0.23 Intentions 0.20 1 0.18 0.09 0.06 0.06 0.07 0.07 0.18 0.10 0.15 0.19 0.06 -0.02 -0.02 0.05 -0.10 HWD2 -0.05 0.23 1 0.38 0.15 0.20 0.37 0.13 0.12 0.07 0.18 0.17 -0.04 -0.01 -0.05 -0.10 -0.14 HWD3 -0.05 0.16 0.51 1 0.15 0.56 0.35 0.08 0.04 -0.03 -0.11 0.02 0.05 0.01 -0.06 -0.13 0.10 HWD4 -0.14 0.01 0.38 0.30 1 0.25 0.25 -0.05 -0.04 0.07 -0.02 -0.10 -0.04 0.06 0.11 0.00 -0.09 HWD5 -0.05 0.09 0.43 0.51 0.36 1 0.51 -0.04 0.04 -0.06 -0.11 0.00 -0.11 -0.10 -0.13 -0.22 -0.05 HWD6 0.05 0.29 0.40 0.53 0.41 0.51 1 0.13 0.19 0.15 0.21 0.25 0.04 0.00 -0.18 -0.27 -0.08 SE2 0.11 -0.06 -0.10 -0.09 0.03 0.00 -0.09 1 0.46 0.50 0.45 0.61 -0.28 -0.28 -0.12 -0.26 -0.11 SE3 0.06 -0.07 -0.08 -0.09 0.14 -0.13 -0.15 0.47 1 0.43 0.44 0.49 -0.11 -0.18 -0.22 -0.17 -0.17 SE4 0.07 -0.05 -0.07 -0.07 0.00 0.04 -0.12 0.48 0.59 1 0.60 0.53 -0.14 -0.13 -0.07 -0.20 -0.03 SE5 0.13 -0.06 -0.04 -0.02 0.03 0.01 -0.01 0.51 0.58 0.54 1 0.63 -0.12 -0.28 -0.19 -0.26 -0.12 SE6 0.09 -0.05 -0.04 0.00 0.11 0.07 -0.01 0.44 0.51 0.59 0.51 1 -0.20 -0.32 -0.36 -0.30 -0.24 TP2 -0.03 0.00 0.02 0.16 -0.01 0.08 0.13 -0.15 -0.31 -0.30 -0.19 -0.28 1 0.38 0.44 0.52 0.44 TP3 -0.17 0.00 -0.04 0.07 -0.19 -0.07 0.05 -0.24 -0.41 -0.36 -0.25 -0.28 0.52 1 0.52 0.57 0.41 TP4 -0.09 -0.12 0.12 0.00 -0.12 -0.02 0.01 -0.06 -0.23 -0.14 -0.20 -0.21 0.47 0.57 1 0.59 0.55 TP5 -0.03 0.09 0.04 0.15 -0.03 0.03 0.13 -0.21 -0.28 -0.22 -0.27 -0.25 0.56 0.54 0.45 1 0.37 TP6 -0.16 -0.08 -0.05 0.10 -0.12 -0.06 0.04 -0.25 -0.32 -0.31 -0.24 -0.25 0.51 0.60 0.61 0.55 1 Mean - 1.35 1.97 1.97 1.51 2.94 2.13 1.88 1.88 1.87 1.80 1.84 1.21 1.26 1.32 1.33 1.25 SD - 1.31 2.55 2.54 2.15 3.42 2.90 0.59 0.61 0.61 0.60 0.63 0.60 0.60 0.68 0.67 0.65 Note. Lower half = ALDH2(-), upper half = ALDH2(+); Feedback = feedback group, Intent = intentions to change drinking behavior, HWD = heaviest week drinking quantity, SE = self-efficacy, TP = threat perception Table 4 Robust Multiple Regression Models of the Effect of Predictors on Post-Feedback Intentions to Change Drinking Behavior in the Full Sample and in Drinkers Only, Including Stratified by ALDH2*2 Full sample Drinkers only All participants n = 280 ALDH2*2(-) n = 147 ALDH2*2(+) n = 133 All drinkers n = 221 ALDH2*2(-) n = 117 ALDH2*2(+) n = 104 β (SE), p β (SE), p β (SE), p β (SE), p β (SE), p β (SE), p Intercept 0.09 (1.359), .474 -0.77 (2.082), .356 -0.59 (1.891), .377 2.39 (1.624), .071 2.78 (2.315), .116 0.84 (2.222), .353 Sex (Male) -0.15 (0.156), .176 -0.12 (0.220), .294 -0.17 (0.239), .237 -0.17 (0.181), .176 -0.16 (0.246), .262 -0.13 (0.267), .313 Age 0.08 (0.066), .107 0.15 (0.100), .073 0.05 (0.092), .300 -0.02 (0.076), .419 0.00 (0.110), .490 -0.03 (0.103), .380 Baseline drinking 0.06 (0.022), .005 0.04 (0.026), .073 0.10 (0.053), .030 0.05 (0.024), .023 0.03 (0.027), .109 0.08 (0.055), .086 Self-efficacy (SE) -0.49 (0.261), .032 -0.67 (0.378), .038 0.08 (0.391), .422 -0.66 (0.308), .017 -0.97 (0.392), .007 0.05 (0.494), .456 Threat perception (TP) -0.79 (0.355), .014 -0.58 (0.552), .147 -0.58 (0.487), .119 -1.25 (0.439), .002 -1.17 (0.577), .022 -1.21 (0.658), .034 Self-efficacy X Threat Perception 0.46 (0.181), .006 0.35 (0.280), .105 0.32 (0.251), .099 0.70 (0.223), .001 0.61 (0.289), .019 0.65 (0.341), .028 ALDH2*2(+) -0.17 (0.265), .258 - - -0.36 (0.299), .116 - - PHEN 0.67 (0.259), .005 0.72 (0.268), .004 0.88 (0.277), .001 0.70 (0.299), .010 0.74 (0.305), .008 1.14 (0.298), .000 PHEN+GENE 0.63 (0.251), .007 0.66 (0.259), .006 0.89 (0.281), .001 0.71 (0.275), .005 0.72 (0.279), .005 1.35 (0.298), .000 ALDH2*2(+)*PHEN 0.19 (0.371), .309 - - 0.28 (0.428), .255 - - ALDH2*2(+)*PHEN+GENE 0.25 (0.371), .250 - - 0.56 (0.416), .091 - - Table 5 Robust Multiple Regression Models of the Effect of Predictors on Post-Feedback Intentions to Change Drinking Behavior in the Full Sample and Split by Low and High Threat Perception Full sample Drinkers only Full sample Low threat High threat Full sample Low threat High threat β (SE), p n = 280 β (SE), p n = 189 β (SE), p n = 91 β (SE), p n = 221 β (SE), p n = 148 β (SE), p n = 73 Intercept 0.09 (1.359), .474 -0.29 (1.607), .427 -2.61 (2.173), .115 2.39 (1.624), .071 1.28 (1.802), .240 -1.00 (2.740), .358 Sex (Male) -0.15 (0.156), .176 -0.10 (0.193), .311 -0.25 (0.316), .218 -0.17 (0.181), .176 -0.16 (0.233), .244 -0.29 (0.339), .197 Age 0.08 (0.066), .107 0.05 (0.085), .269 0.16 (0.117), .091 -0.02 (0.076), .419 -0.03 (0.094), .377 0.05 (0.151), .359 Baseline drinking 0.06 (0.022), .005 0.07 (0.032), .015 0.04 (0.037), .163 0.05 (0.024), .023 0.04 (0.035), .113 0.04 (0.038), .155 Self-efficacy (SE) -0.49 (0.261), .032 -0.10 (0.160), .265 0.60 (0.267), .013 -0.66 (0.308), .017 -0.04 (0.190), .416 0.79 (0.305), .005 Threat perception (TP) -0.79 (0.355), .014 - - -1.25 (0.439), .002 - - Self-efficacy X Threat Perception 0.46 (0.181), .006 - - 0.70 (0.223), .001 - - ALDH2*2(+) -0.17 (0.265), .258 -0.04 (0.340), .458 -0.58 (0.487), .118 -0.36 (0.299), .116 -0.13 (0.386), .364 -0.65 (0.538), .113 PHEN 0.67 (0.259), .005 0.73 (0.314), .010 0.56 (0.539), .151 0.70 (0.299), .010 0.84 (0.364), .011 0.53 (0.615), .194 PHEN+GENE 0.63 (0.251), .007 0.75 (0.319), .009 0.28 (0.468), .277 0.71 (0.275), .005 0.92 (0.358), .005 0.33 (0.485), .251 ALDH2*2(+)*PHEN 0.19 (0.371), .309 0.38 (0.462), .209 -0.13 (0.732), .430 0.28 (0.428), .255 0.37 (0.538), .244 0.05 (0.814), .474 ALDH2*2(+)*PHEN+GENE 0.25 (0.371), .250 0.29 (0.478), .269 0.22 (0.674), .373 0.56 (0.416), .091 0.54 (0.544), .160 0.33 (0.731), .324 Table 6 Robust Multiple Regression Models of the Effect of Predictors on the Change in Heaviest Week Drinking Quantity from Baseline by Study Wave Full Sample Drinkers only ALDH2*2(-) ALDH2*2(+) ALDH2*2(-) ALDH2*2(+) β (SE), p β (SE), p β (SE), p β (SE), p β (SE), p β (SE), p 30-day change n = 277 n = 146 n = 127 n = 217 n = 114 n = 99 Intercept 1.08 (1.406), .221 0.97 (2.535), .351 0.88 (1.423), .268 2.97 (2.450), .113 3.85 (4.187), .180 2.72 (2.833), .169 Sex (Male) -0.15 (0.180), .208 -0.20 (0.298), .253 -0.09 (.201), 0.325 -0.08 (0.315), .405 -0.23 (0.508), .323 0.10 (0.400), .399 Age -0.03 (0.075), .348 -0.02 (0.136), .435 -0.03 (.077), 0.354 -0.11 (0.130), .196 -0.16 (0.223), .243 -0.10 (0.151), .249 Baseline drinking -0.32 (0.026), .000 -0.32 (0.036), .000 -0.28 (.045), .000 -0.39 (0.042), .000 -0.38 (0.056), .000 -0.41 (0.082), .000 Intentions to change 0.10 (0.070), .087 0.14 (0.118), .112 0.05 (0.074), .266 0.10 (0.123), .204 0.12 (0.210), .284 0.08 (0.141), .281 ALDH2*2(+) -0.08 (0.308), .393 - - -0.02 (0.518), .486 - - PHEN -0.32 (0.300), .144 -0.34 (0.373), .179 -0.11 (0.243), .330 0.04 (0.524), .469 0.09 (0.642), .445 -0.09 (0.459), .419 PHEN+GENE 0.05 (0.289), .433 0.06 (0.360), .434 -0.29 (0.247), .122 0.33 (0.481), .247 0.29 (0.590), .312 -0.57 (0.481), .119 ALDH2*2(+)*PHEN 0.17 (0.429), .347 - - -0.16 (0.739), .412 - - ALDH2*2(+)*PHEN+GENE -0.45 (0.431), .150 - - -0.97 (0.734), .093 - - 4-month change n = 247 n = 130 n = 113 n = 190 n = 98 n = 88 Intercept -4.02 (1.630), .007 -7.83 (3.368), .011 -1.47 (1.241), .119 -4.45 (2.591), .044 -4.78 (5.342), .187 -2.39 (2.023), .121 Sex (Male) -0.32 (0.205), .060 -0.23 (0.381), .278 -0.17 (0.177), .168 -0.28 (0.326), .199 -0.12 (0.627), .423 -0.11 (0.284), .355 Age 0.28 (0.087), .001 0.49 (0.180), .004 0.10 (0.067), .063 0.34 (0.137), .007 0.36 (0.284), .103 0.18 (0.108), .053 Baseline drinking -0.68 (0.029), .000 -0.61 (0.045), .000 -0.86 (0.039), .000 -0.74 (0.042), .000 -0.66 (0.067), .000 -0.90 (0.058), .000 Intentions to change -0.08 (0.079), .169 -0.15 (0.145), .155 -0.01 (0.066), .451 -0.19 (0.125), .069 -0.35 (0.245), .077 -0.03 (0.102), .372 ALDH2*2(+) -0.72 (0.349), .021 - - -1.06 (0.526), .023 - - PHEN -0.32 (0.339), .174 -0.28 (0.470), .278 -0.1 (0.216), .330 0.50 (0.544), .180 0.75 (0.786), .172 -0.28 (0.327), .196 PHEN+GENE -0.49 (0.335), .073 -0.32 (0.465), .246 -0.18 (0.217), .203 -0.42 (0.499), .200 -0.30 (0.719), .338 -0.24 (0.342), .239 ALDH2*2(+)*PHEN 0.02 (0.485), .483 - - -0.75 (0.752), .160 - - ALDH2*2(+)*PHEN+GENE 0.24 (0.495), .315 - - 0.44 (0.753), .281 - - 7-month change n = 257 n = 140 n = 113 n = 198 n = 107 n = 87 Intercept 6.87 (2.836), .008 9.81 (4.826), .022 4.68 (2.599), .037 14.31 (4.12), .000 20.16 (6.68), .002 12.58 (4.67), .004 Sex (Male) -0.41 (0.352), .121 -0.76 (0.556), .088 -0.06 (0.356), .439 -0.41 (0.506), .211 -1.23 (0.795), .062 0.78 (0.615), .105 Age -0.27 (0.152), .039 -0.42 (0.259), .054 -0.18 (0.141), .098 -0.61 (0.219), .003 -0.90 (0.356), .007 -0.56 (0.251), .014 Baseline drinking -0.32 (0.052), .000 -0.27 (0.069), .000 -0.53 (0.083), .000 -0.43 (0.07), .000 -0.35 (0.091), .000 -0.74 (0.133), .000 Intentions to change -0.09 (0.134), .256 -0.08 (0.209), .343 -0.08 (0.135), .285 -0.32 (0.193), .051 -0.47 (0.308), .064 -0.20 (0.223), .187 ALDH2*2(+) -0.48 (0.647), .230 - - -0.73 (0.886), .207 - - PHEN -0.59 (0.582), .154 -0.46 (0.691), .254 -0.26 (0.455), .288 0.43 (0.823), .300 0.70 (0.983), .238 0.17 (0.759), .412 PHEN+GENE -0.14 (0.575), .403 -0.23 (0.681), .367 -0.43 (0.458), .177 0.15 (0.778), .424 -0.06 (0.925), .472 0.11 (0.781), .446 ALDH2*2(+)*PHEN 0.49 (0.853), .284 - - -0.10 (1.192), .465 - - ALDH2*2(+)*PHEN+GENE -0.18 (0.873), .420 - - -0.06 (1.210), .481 - - 10-month change n = 246 n = 131 n = 111 n = 189 n = 100 n = 85 Intercept 2.32 (2.128), .138 0.84 (4.351), .424 2.32 (2.166), .143 4.94 (3.681), .091 4.08 (7.063), .282 6.01 (3.702), .054 Sex (Male) 0.08 (0.264), .374 0.32 (0.479), .251 -0.03 (.302), 0.466 0.56 (0.455), .108 0.72 (0.793), .184 0.70 (0.506), .084 Age -0.10 (0.113), .198 -0.03 (0.233), .453 -0.09 (0.117), .210 -0.21 (0.195), .146 -0.17 (0.375), .323 -0.26 (0.197), .093 Baseline drinking -0.39 (0.039), .000 -0.35 (0.057), .000 -0.55 (0.082), .000 -0.50 (0.062), .000 -0.46 (0.086), .000 -0.72 (0.128), .000 Intentions to change 0.15 (0.104), .079 0.33 (0.187), .039 0.03 (0.116), .408 0.07 (0.181), .354 0.27 (0.322), .206 -0.07 (0.191), .363 ALDH2*2(+) -0.13 (0.478), .389 - - -0.26 (0.788), .371 - - PHEN 0.19 (0.455), .341 0.15 (0.609), .402 -0.01 (0.374), .488 1.25 (0.775), .054 1.11 (1.013), .138 0.18 (0.608), .383 PHEN+GENE 0.00 (0.436), .496 -0.08 (0.582), .447 0.08 (0.381), .416 0.37 (0.704), .298 0.27 (0.914), .385 0.33 (0.638), .305 ALDH2*2(+)*PHEN -0.27 (0.646), .338 - - -1.21 (1.092), .135 - - ALDH2*2(+)*PHEN+GENE -0.06 (0.649), .465 - - -0.29 (1.082), .395 - - Table 7 Full Path Model Parameter Estimates with Robust Standard Errors ALDH2*2(-) ALDH2*2(+) From To β (SE), p β (SE), p Threat perception PHEN -0.11 (0.062), .067 0.02 (0.062), .731 Threat perception PHEN+GENE 0.05 (0.060), .420 -0.06 (0.064), .353 Self-efficacy PHEN 0.03 (0.061), .602 -0.03 (0.067), .612 Self-efficacy PHEN+GENE 0.05 (0.059), .406 -0.03 (0.066), .703 Threat perception*Self-efficacy PHEN -0.04 (0.090), .626 -0.06 (0.088), .512 Threat perception*Self-efficacy PHEN+GENE 0.11 (0.097), .242 0.14 (0.085), .096 PHEN Intentions to change 0.70 (0.235), .003 1.01 (0.237), .000 PHEN+GENE Intentions to change 0.71 (0.209), .001 0.91 (0.226), .000 Threat perception Intentions to change 0.11 (0.168), .511 0.17 (0.175), .323 Intentions to change Change in drinking 0.34 (0.195), .083 0.14 (0.151), .345 Baseline drinking Change in drinking -0.42 (0.117), .000 -0.37 (0.117), .002 Age Intentions to change 0.13 (0.094), .168 0.12 (0.084), .164 Sex Intentions to change -0.33 (0.192), .090 0.00 (0.204), .982 Fit Statistics # of Parameters 58 RMSEA 0.107 AIC 1497 BIC -8393 Note. Threat perception and self-efficacy were obtained pre-feedback at wave 2 Table 8 Full Path Model Parameter Estimates for Drinker-Only Sample with Robust Standard Errors ALDH2*2(-) ALDH2*2(+) From To β (SE), p β (SE), p Threat perception PHEN -0.12 (0.066), .066 0.04 (0.069), .578 Threat perception PHEN+GENE 0.03 (0.072), .665 -0.02 (0.071), .726 Self-efficacy PHEN 0.01 (0.063), .849 -0.03 (0.073), .726 Self-efficacy PHEN+GENE 0.07 (0.065), .270 -0.05 (0.071), .521 Threat perception*Self-efficacy PHEN 0.00 (0.090), .985 -0.02 (0.097), .850 Threat perception*Self-efficacy PHEN+GENE 0.08 (0.106), .423 0.12 (0.100), .244 PHEN Intentions to change 0.90 (0.261), .001 1.09 (0.256), .000 PHEN+GENE Intentions to change 0.79 (0.229), .001 1.25 (0.257), .000 Threat perception Intentions to change 0.05 (0.176), .767 0.19 (0.216), .384 Intentions to change Change in drinking 0.40 (0.258), .117 0.12 (0.192), .539 Baseline drinking Change in drinking -0.46 (0.128), .000 -0.43 (0.129), .001 Age Intentions to change 0.01 (0.099), .916 0.05 (0.097), .616 Sex Intentions to change -0.41 (0.204), .046 -0.01 (0.234), .949 Fit Statistics # of Parameters 58 RMSEA 0.106 AIC 1339 BIC -5443 Note. Threat perception and self-efficacy were obtained pre-feedback at wave 2 Figure 1 Multigroup Path Model Figure 2 5% trimmed mean trajectories of heaviest drinking week quantity split by ALDH2*2 status and feedback group Figure 3 Fitted Multigroup Path Model Figure 4 Fitted Multigroup Path Model for the Drinker-Only Sample

Abstract (if available)

Abstract Both flushing and the genetic variant ALDH2*2 are known risk factors for alcohol-related cancers within the research community, but the general population is largely unaware of these risks. This creates potential for brief educational interventions to increase awareness and reduce the alcohol-related cancer burden. Current findings on the degree to which personalized health feedback motivates behavior change are mixed, however. Individual differences like self-efficacy and threat perception are established mechanisms of health behavior change and may explain some of the discrepancies in how people respond to knowledge about health risks. In the current study, we examine the effects of a brief, online feedback intervention about flushing, ALDH2*2, alcohol use, and cancer risk on participants’ intentions to change their drinking behavior and on their actual heaviest week drinking behavior over the following year. We also evaluate the degree to which self-efficacy and threat perception relate to participants’ intentions to change their drinking behavior. Results showed that receiving risk feedback increased participants’ intentions to change their drinking regardless of their ALDH2*2 status. Self-efficacy and threat perception were also related to participants’ intentions to change their drinking behavior. These findings suggest that learning about the cancer risk related to alcohol use affected participants’ intentions to change their drinking behavior, but this, in turn, did not relate to significant changes in their heaviest week drinking quantity in the 10 months post-feedback.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Personalized normative feedback applied to undergraduates with problem drinking: a comparison with psychoeducation and an examination of cognitive-affective change mechanisms via the articulated ...

PDF

Breaking it down to make it stronger: examining the role of source credibility and reference group specificity in the influence of personalized normative feedback on perceived alcohol use norms a...

PDF

Exploring experiences of minority-stress and alcohol use among sexual minorities using articulated thoughts in simulated situations: a proof of concept study

PDF

The tracking effect: tracking and the impact on self-efficacy in middle school students

Asset Metadata

Creator Saldich, Emily Briggs (author)

Core Title Examining the effects of personalized feedback about ALDH2*2, alcohol use, and associated health risks on drinking intentions and consumption: the role of self-efficacy and perceived threat

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Psychology

Degree Conferral Date 2024-12

Publication Date 01/12/2025

Defense Date 04/22/2024

Publisher Los Angeles, California (original), University of Southern California (original), University of Southern California. Libraries (digital)

Tag alcohol metabolism,aldehyde dehydrogenase,brief intervention,cancer risk,Flushing,OAI-PMH Harvest,perceived threat,self-efficacy

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Luczak, Susan (committee chair), Beam, Christopher (committee member), Lai, Mark (committee member)

Creator Email esaldich@gmail.com,saldich@usc.edu

Unique identifier UC11399FAEH

Identifier etd-SaldichEmi-13742.pdf (filename)

Legacy Identifier etd-SaldichEmi-13742

Document Type Thesis

Format theses (aat)

Rights Saldich, Emily Briggs

Internet Media Type application/pdf

Type texts

Source 20250113-usctheses-batch-1234 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu