GENDER DIFFERENCES IN THE ILLICIT USE OF PRESCRIPTION PAIN MEDICATION. Sean Esteban1, Carol Boyd2, Christian J. Teter3, 1University of Michigan Substance Abuse Research Ctr, Ann Arbor, MI United States; 2University of Michigan, Ann Arbor, MI United States; 3Northeastern Univ., Bouve College of Health Sciences, Belmont, MA United States
The illicit use of prescription pain medications among adolescents and college students in the United States represents a growing public health problem (e.g. Johnston et al., 2004, SAMHSA, 2004). Indeed, the illicit use of prescription pain medication has increased significantly among college students (e.g., Johnston et al., 2003a; Mohler-Kuo et al., 2003). Despite the high prevalence and recent increases in illicit use, prescription pain medications remain a highly effective and safe medication for the majority of patients suffering from acute and chronic pain. Therefore, there is need to strike a balance between the medical necessity to treat patients with prescription pain medications and the need to reduce illicit use of these abusable drugs (Joranson et al., 2000; Ling et al., 2003; Simoni-Wastila and Tompkins, 2001). Little epidemiological research exists that accurately assesses the characteristics of those individuals most at risk for illicit use of prescription pain medication (Zacny et al., 2003).
This study examines the gender differences in the prevalence, correlates, and sources associated with the illicit use of prescription pain medication (e.g., hydrocodone, oxycodone, codeine) among undergraduate college students. A cross-sectional, Web-based survey was self-administered by a large random sample of 9,161 undergraduate students attending a large Midwestern university in the United States. Although undergraduate women were more likely to be prescribed pain medication, men were more likely to be approached to divert their prescription pain medication and report illicit use of prescription pain medication. Multivariate analyses indicated past year illicit use of prescription pain medication was higher among undergraduate students who were: previously prescribed pain medication, living in a house or apartment, and earning lower grade point averages. The majority of students obtained prescription pain medication for illicit use from peers and the second leading source was family members. There were several gender differences in the risk factors and sources of illicit use of prescription pain medication. Based on qualitative data, illicit use included recreational use for the purposes of intoxication and self-medication for pain episodes. Illicit users who obtained prescription pain medication from peers reported significantly higher rates of other substance use while those who obtained prescription medications from family members did not. These findings suggest that the illicit use of prescription pain medications may represent a problem among undergraduate students and effective prevention efforts are needed that account for gender differences.
ABUSE OF PRESCRIPTION PAIN MEDICATION BY ADOLESCENTS: RESULTS OF A WEB-BASED SURVEY. Carol Boyd1, Sean Esteban2, Christian J. Teter3, 1University of Michigan, Ann Arbor, MI United States; 2University of Michigan Substance Abuse Research Ctr, Ann Arbor, MI United States; 3Northeastern Univ., Bouve College of Health Sciences, Belmont, MA United States
There is strong evidence that prescription drugs are being used non-medically by U.S. secondary school students. This form of illicit drug use is second only to marijuana among adolescent users (Johnston, et al, 2004, Office of Applied Studies, 2002) and often involves medications with high addictive potential (e.g. opiate analgesics). And yet, despite this worrisome trend, there is the paucity of data on the characteristics associated with the non-medical use of opiate pain medication among middle and high school students (Zacny, et al., 2003). Thus, the purpose of this descriptive, web-based study was to examine the non-medical use (illicit use) of prescription pain medications in a sample of 5th through 10th graders (N=1,017) residing in one school district in metropolitan Detroit.To meet the research aims, we examined the prevalence of medical and non-medical use of prescription pain medication. Twenty-two percent of girls and ten percent of boys reported lifetime, non-medical use of a pain medication (χ2 = 25.2, df = 1, p < 0.001); fifteen percent of girls and seven percent of boys reported past year use (χ2 = 12.7, df = 1, p < 0.01). We further investigated the substance-abuse characteristics of four, mutually exclusive groups: 1) non-users of prescription pain medications; 2) medical (licit) users of prescription pain medications; 3) non-medical (illicit) users of prescription pain medications and 4) users of both medical and non-medical prescription pain medications (licit and illicit users). There were notable group differences: When compared to non-users, those adolescents reporting non-medical pain medication use (group 4) were eight times more likely to smoke cigarettes, seven times more likely to drink and smoke marijuana, four times more likely to binge drink and 10 times more likely to have used several other drugs. Adolescents who reported both medical and non-medical pain medication use (group 3) were also more likely to use cigarettes, alcohol, marijuana and other illicit drugs. Students who only used pain medication as prescribed (group 2) were significantly more likely to use cigarettes and marijuana but no more likely to use alcohol or other illicit drugs. We conclude this paper by noting that if the prevention community is to combat the non-medical use of prescription pain medications, better research is needed to characterize those most at risk.
CONCURRENT 9, METHODS, Organized Symposia
ADVANCES IN LATENT VARIABLE MODELING
Chair: Daniel Feaster
ADVANCES IN LATENT VARIABLE MODELING. Daniel Feaster1, Bengt Muthen2, Karen Nylund2, 1University of Miami, Miami, FL United States; 2University of California, Los Angeles, Los Angeles, CA United States
This symposium presents new methodology for analyzing intervention impact with latent variable models. A number of advances in this methodology have taken place in recent years, providing researchers with a broad array of tools to analyze the impact of an intervention over time. Many of these methods relate to examining variation in impact of an intervention. These range from completely parametric models that examine treatment by baseline interactions on growth parameters to those involving nonparametric mixture modeling, as in the general growth mixture model (GGMM). We explore a number of advances in this methodology and present new results of analyses on existing preventive field trials. These new methods include the use of nonlinear transformations of data and nonlinear additive model fits, and interactions involving multiple latent variables or higher order interactions of latent variables. We also describe new methods for handling missing data and participation bias in intervention trials.
POWER TRANSFORMATIONS AND ADDITIVE MODELS IN MODELING INTERVENTION IMPACT WITH GROWTH MODELS. C. Hendricks Brown1, Wei Wang1, 1University of South Florida, Tampa, FL United States
In population-based trials, a common finding has been that the preventive intervention´s effect varies by person-level and context-level risk status. Furthermore, the impact on early developmental course may affect more distal outcomes such as diagnoses in later stages of life. This presentation discusses new methodologic approaches to characterize the impact on both proximal growth and distal diagnoses. Treatment by baseline interaction growth models provide one way to examine variation in impact that is linear on growth. Another method is General Growth Mixture Models (GGMM) which incorporates discrete classes of growth trajectories. In this presentation we incorporate two new flexible techniques, providing much more flexible fits than we had previously. The first involves nonlinear transformation of the outcome measures. We present methods that allow for comparisons of the likelihoods under different monotone transformations. Secondly, we introduce additive models into the latent variable framework underlying these growth models. To date such nonparametric patterns of growth could not be modeled satisfactorily. We present their theoretical development and give examples of this methodology using the Baltimore Prevention Program first generation trial, which aims at reducing aggression and its long-term consequences. Our methods provide substantial improvements of fit compared to existing methods.
MODELING HETEROGENEITY OF GROWTH AND EVALUATING FIT IN A NONLINEAR SEM FRAMEWORK. Andreas Klein1, Bengt Muthen2, 1University of Illinois at Urbana-Champaign, Champaign, IL United States; 2University of California, Los Angeles, Los Angeles, CA United States
A general latent variable modeling framework and estimation method is presented which incorporates the modeling of multiple nonlinear effects in both longitudinal and cross-sectional models. For the case of longitudinal models, the applicability of the framework is illustrated by a heterogeneous growth curve model. Formally, heterogeneity of growth refers to the fact that some subgroups of individuals grow more consistently than others, and to model this heterogeneity statistically correct is essential for optimal prediction of individuals´ development. The new methodology enables the researcher to analyze more efficiently what heterogeneity of growth could depend on and provides a tool for an improved prediction of individual growth, based on intial status or other covariate information. The methodology is compared to the popular technique of mixed growth curve modeling. For the case of cross-sectional models, the framework includes models with multiple interaction effects among latent exogenous variables. In particular, the question of model fit will be discussed for nonlinear SEM and several ways to evaluate model fit will be suggested. The proposed model, estimation technique, and model procedures for model evaluation will be explained using an empirical example.
RANDOMIZED TRIALS WITH NON-PARTICIPATION AND NON-RESPONSE: MODEL CHOICES IN INTENTION-TO-TREAT ANALYSIS. Booil Jo1, Elizabeth Ginexi2, George Howe3, 1Stanford University, Stanford, CA United States; 2National Institutes of Health, Bethesda, MD United States; 3George Washington University, Washington, DC United States
Randomized trials often suffer from both non-participation and missingness in outcomes due to non-response. Due to the similar nature of non-participation in treatment and non-response at follow-ups, it is often suspected that participation behavior may be connected to response patterns. Focusing on the possible association between these two, this study considers four scenarios of missing data mechanisms. They are 1) non-response is not associated with non-participation, 2) non-response is associated with non-participation among individuals with observed participation information, 3) non-response is associated with non-participation not only among individuals with observed participation information, but also among individuals without observed participation information, and 4) non-response is associated with other unobserved variables than participation status. Given these four conditions, three ways of comparing treatment groups will be discussed. They are a) standard intent-to-treat (ITT) analysis assuming that 1) is true, b) ITT analysis based on the instrumental variable (IV) approach assuming that 2) is true, and c) ITT analysis based on IV approach assuming that 3) is true. Previous studies (Frangakis & Rubin, 1999; Peng, Little, & Raghunathan, 2004) compared some of these methods based on simulation studies, but their results did not provide enough information for applied researchers to understand their analysis choices. This study first carefully compares these alternative models using an analytical approach. Practical guidelines for model choices are then discussed using the data from Couples Employment Project (University of Michigan's Institute for Social Research and the Center for Family Research at the George Washington University), a randomized field experiment designed to prevent depression and to promote positive interactions among spouses and committed partners.
5:00 PM – 7:00 PM
POSTER SESSION 1, TECHNOLOGY DEMONSTRATIONS AND RECEPTION
Posters are grouped by theme
POSTERS: CULTURAL SENSITIVITY
TYPES OF SPOUSAL ABUSE: THEIR CORRELATES IN AN AFRICAN CULTURE. Magen Mutepfa1, 1Schools Psychological Services (Zimbabwe), Harare, Zimbabwe Zimbabwe
The study investigated physical, emotional, economical, sexual and psychological spousal abuse in a sample of Zimbabwean workers. A total of 270 people took part in this study (Males= 158; Females = 106; Mean age 33.8 years: SD=6.8 years). Psychological, emotional and economic abuse was positively and significantly correlated. Physical abuse was not significantly correlated to any of the other three types of abuse. The relationships held regardless of gender, social class or religiosity. The view that physical spousal abuse was more likely among those with lower socio-economic status, females and non-religious groups was not supported.
A CULTURALLY SENSITIVE MODEL FOR REDUCINGYOUTHS' RISK FOR VIOLENCE: FROM THEORY TO PRACTICE. Paulette Hines1, William Vega, Ph.D.1, John Jemott2, 1University of Medicine and Dentistry of New Jersey, Piscataway, NJ United States; 2University of Pennsylvania, Phildelphia, PA United States
This interactive session will involve an overview of SANKOFA – a theory-based, research validated violence prevention program specifically designed for urban minority youth. The presenters will review the results of an NIH-sponsored controlled trial using the intervention model in New Jersey. The program´s goal is to equip youths with the tools –knowledge, attitudes, confidence, motivation, environmental context, and actual skills –essential to promoting resilience and minimizing risk for violence perpetration (including bullying), violence victimization, and other maladaptive behaviors. This multi-tiered program, manualized to facilitate replication with fidelity, includes training for adolescents, parents, educators and other youth service providers. The program is currently being disseminated through schools, youth detention and juvenile justice systems, as well as faith-based and community-based organizations. The presenters will provide an overview of (1) the rationale, theory, and objectives that undergird the SANKOFA program; (2) key issues in designing the program curricula and intervention strategies; (3) implementation considerations; and (4) research outcomes.
The outcome analysis demonstrates that the intervention was effective in reducing a number of risk factors and violence-related outcomes among high school youths, especially as contrasted with controls. This intervention trial was particularly interesting because it was conducted in two intervention locations in New Jersey, characterized as high risk and moderate risk environments, where the youths are predominantly African American and Latino. A second topic of interest concerns the generalizability of culturally-tailored interventions for use with youths of other ethnic backgrounds. SANKOFA was designed originally as an Afrocentric intervention model. As occurs in many urban school situations, there are also many Latino students in the same urban areas and high schools within New Jersey. For both scientific and pragmatic reasons, these youths were also included in the intervention trial and we will report on both the outcomes and experience of this multi-ethnic inclusion in the intervention trial.
The presenters will address the workshop objectives via lecture, visuals, group discussion, and handouts. Participants will also have the opportunity to observe the program in action via videotaped excerpts, time permitting.
MODEL MINORITY – EMPIRICALLY VALID OR JUST A STEREOTYPE? EXAMINATION OF THE PREVALENCE AND DISTRIBUTION OF YOUTH PROBLEM BEHAVIORS AMONG ASIAN AMERICAN AND SUBGROUPS. Yoonsun Choi1, 1University of Chicago, Chicago, IL United States
U.S. has seen a rapid growth in Asian American families in recent years, but little is known about the mental health of Asian American youth. We do not have reliable knowledge on how prevalent problem behaviors are among these youth and how those behaviors change over time. Asian Americans are stereotypically regarded as the “model minority”. There is little empirical and reliable data to evaluate the stereotype. If, however, the stereotype of the “model minority” is valid, it will be beneficial to identify the factors that yield these improved outcomes and enhance these factors to buffer risk of problems. In contrast to this stereotype, the few studies that exist on Asian youth in America present a fragmented and bimodal set of findings. Some studies report positive and healthy outcomes among Asian youth, like higher levels of academic achievement and aspiration and lower rates of problem behaviors. At the same time, others report various problems like an upsurge in Asian gangs and violent crimes, arguing that Asian youth are vulnerable to multiple risks to the same extent as, if not more than, other groups of youth. The methodological limitations of most of the studies of Asian American youth to date, such as the regional nature of studies and the varying measures used, have enabled this mixed picture of Asian youth to surface from these studies. It is possible, however, that there actually is a bimodal distribution of developmental outcomes for Asian youth in U.S., with both notable success and failure. If the latter is true, a critical examination of the Asian youth development can illuminate the factors that promote divergent outcomes. Using existing data from the National Longitudinal Study of Adolescent Health (Add Health), this study examines the prevalence of problem behaviors among Asian American adolescents and compares it to that of European American, African Americans, Hispanic and multiracial youth. Comparisons are also made among subgroups of Asian American youth (e.g. Chinese, Filipino, Koreans, and Vietnamese). The use of national samples helps avoid the sampling problems posed by many previous studies. In addition, this study explicitly tests the distribution of problem behaviors. For example, the residuals of the outcomes adjusted for age, gender, SES, and immigrant status are compared across groups to detect a bimodal distribution among Asian American youth.
ADHD SYMPTOMATOLOGY & HARSH PARENTING PRACTICES. Laura Harver1, Elizabeth MacKenzie2, Martie Skinner1, Kevin Haggerty1, 1University of Washington, Seattle, WA United States; 2Social Development Research Group, Univ. of Washington, Seattle, WA United States
The current study was conducted as part of the NIDA summer internship for underrepresented minority undergraduate students. We explored the relationship between attention problem behaviors, parenting and race. Attention problems and other symptoms of Attention Deficit/Hyperactivity Disorder put children at risk for negative youth outcomes such as substance abuse, delinquency, and depression (Barkley, 1997, 2000; Root & Resnick, 2003). In contrast to other disruptive behavior problems, the etiology of ADHD symptomatology is believed to be related more strongly to individual child factors (e.g. neurobiological and genetic factors) than to parenting factors (Biederman & Faraone, 2002; Root & Resnick, 2003). Attention problems are correlated with harsh parenting (Lindahl, 1998) and it is posited that this relationship is at least partly explained as a parenting response to the increased child-rearing challenges posed by attention problems (Dacey & Travers, 2004). We examined the relationship between parent-reports of child attention problems and harsh parenting in a sample of 331 families with young adolescents. The sample was balanced for race (n = 162 African American, and n = 169 European American), which allowed us to examine whether the relationship between child attention problems and harsh parenting is moderated by race. Measures were collected from both parents and adolescents. We hypothesized a positive relationship between attention problems and harsh parenting. We further hypothesized that the relationship between child attention problems and harsh parenting and would be moderated by race. Specifically, we predicted a positive relationship between the variables for European American, but not African American families. Results of Analysis of Variance (ANOVA) procedures revealed a main effect for attention problems on the prediction of harsh parenting; attention problems were positively related to harsh parenting. There was also a moderating effect for race. The relationship between attention problems and harsh parenting were significantly stronger for European American families than for African American families. Implications for family research and preventive intervention are discussed.
DIFFERENTIAL ITEM FUNCTIONING OF AFRICAN AMERICANS AND CAUCASIANS IN THE BECK DEPRESSION INVENTORY. Sarah Galloway1, Tracy Kline1, Melvin Wilson1, Daniel Shaw2, 1University of Virginia, Charlottesville, VA United States; 2University of Pittsburgh, Pittsburgh, PA United States
Depressive disorders can be extremely disabling and associated with a low level of quality of life. Research has indicated that the prevalence of depression for African Americans (AA) is similar to that of Caucasians (CA) with 3.1% of AAs presenting symptoms of Major Depressive Disorder (Jackson-Triche et al., 2001). The current study investigates data collected from the Early Steps Project (ES), a family-centered intervention focusing on preventing the onset and severity of disordered conduct in children of low-income families. An important aspect of the ES prevention involves implementing a culturally sensitive approach with an ethnically diverse sample. Consequently, it is critical to evaluate the ES sample for cultural patterns in response performance. For this analysis, the participants consist of 85 mothers—45 AAs and 40 CAs.
The Beck Depression Inventory (BDI), a commonly used self-report measure, assists in detecting symptoms and possible severity of depression. The BDI is comprised of 21 items in which respondents rate on a scale from zero to three intensity of symptoms and attitudes congruent with depression. Beck, Steer and Garbin (1988) recommend a three-factor structure solution of BDI: Negative Attitudes Toward Self, Performance Impairment, and Somatic Disturbance.
Using a Differential Item Function (DIF), the BDI was analyzed to examine how CA and AA females endorse response categories. Preliminary results show no overall differences in ethnic group depression scores (p=.25), but differences exist at the item level. CAs report stronger feelings of being punished (p=.01), unattractive (p=.1), and tired (p=.05) than AAs. Additionally, they report suicide ideation in which they would not carry out while AAs are more likely to say that they do not have suicidal thoughts (p=.1). On the other hand, AAs state a greater loss in sexual interest (p=.1) and degree of self blame for negative occurances (p=.1) than CAs.
Although BDI means are similar for AAs and CAs, initial findings indicate they that report symptoms of depression differently. This suggests that the factors are not accurate descriptions of the expression of depression. An understanding of item level responses and scale utilization is essential when using the BDI as a screening and diagnostic tool. The results will be discussed in terms of the influence of ethnically specific response patterns on screening, diagnosing and treating a diverse sample population.