PRESCHOOL LANGUAGE-FOCUSED CURRICULA: PREVENTING ACADEMIC PROBLEMS THROUGH EARLY INTERVENTION. Laura Justice1, Alice Wiggins1, Khara Pence1, 1University of Virginia, Charlottesville, VA United States
This paper presents theoretical and design considerations in early childhood prevention research, particularly as they apply to preventing academic failure through delivery of high-quality preschool curricula for children of poverty. A pressing problem in education sciences is the need for evidence-based primary prevention practices that accelerate early language development and reduce the occurrence of school-age language impairment and related problems (reading disability, academic failure). The need for primary prevention delivered early in children´s life is suggested by data showing that once oral language problems reach a level of clinical significance, a reversal of competence to normal levels of functioning is highly unlikely (Johnson et al., 1999). Moreover, a substantial research corpus shows strong integrative relationships between preschool language deficits and later reading disabilities (e.g., Catts, Fey, Zhang, & Tomblin, 2001).
There is a clear need for establishing a broad evidentiary base for primary prevention strategies targeting oral language that may then be universally applied to the more than 5 million children of poverty in this country. In this paper, we provide preliminary findings from implementing a comprehensive language-focused preschool prevention curriculum for 4-year-old children from low-income households. The aim of the curriculum is to bring children´s language achievements to average (or higher) levels of performance by the end of preschool and for children to maintain these gains through first grade. In 2003-2004, we randomly assigned 14 preschool classrooms serving 198 4-year-old at-risk children to the experimental curriculum or to maintain their prevailing curriculum to determine the extent to which a language-focused curriculum produces educationally meaningful short- and long-term effects. While the first reports of impact will not be available until the summer of 2005 from the U. S. Department of Education* , in this paper we provide theoretical bases for preventing language difficulties in young children, design consideration in implementing preschool language-focused curricula, and preliminary findings concerning individual differences in children´s rate of language growth within the curriculum.
*The findings reported here are based on research conducted by the authors as part of the Preschool Curriculum Evaluation Research (PCER) program funded by the Institute of Education Sciences (IES), U.S. Department of Education .The PCER Consortium consists of representatives from IES, the national evaluation contractors, Research Triangle Institute (RTI) and Mathematica Policy Research, Inc. (MPR), and each grant site participating in the evaluation.
PROMOTING SOCIAL AND ACADEMIC COMPETENCE IN THE CLASSROOM: AN INTERVENTION STUDY OF THE RESPONSIVE CLASSROOM APPROACH.. Sara Rimm-Kaufman1, Iris Chiu1, 1University of Virginia, Charlottesville, VA United States
A growing literature points to the importance of classroom social processes for mitigating risk and improving children´s competence, suggesting the critical role that teachers play in creating a classroom environment conducive to learning. The present paper examines the effectiveness of the Responsive Classroom (RC) Approach. This intervention recommends principles and practices to integrate social and academic learning, address discipline problems proactively, and enhance children´s self-regulation and cooperation.
Two research questions are addressed in this two-year longitudinal, quasi-experimental study. 1) How does teachers´ use of RC practices contribute to children´s academic and social growth? 2) Is the relation between teachers´ use of RC practices and children´s growth moderated by children´s sociodemographic risk?
Participants were 157 children (71 girls, 59 “at risk” for school failure) and 62 teachers (years experience 1-34, mean=11.0) in grades K-4 at six schools (3 RC, 3 comparison). In year 1, families reported family income, home language, mother education, and marital status as risk indicators. Teachers (grades K-3) reported on academic skills, their relationship with the study child, social skills, and social competence.
In year 2, children´s current teachers (grades 1-4) completed these same questionnaires. Also, in year 2, teachers reported on their use of RC classroom practices using a 39-item questionnaire. (This questionnaire was highly correlated [r =.70] to observed RC classroom practices in a validity study of 68 teachers.)
Fourteen regression analyses were computed. For each, a year 1 academic or social skills variable was treated as a covariate; risk status (no risk vs. one or more factor) and teachers´ report of RC practices were independent variables; and children´s year 2 academic and social skills were dependent variables.
Results showed a positive relation between RC practices and growth in reading performance, teachers´ perception of closeness toward the child, and prosocial behavior with peers. An inverse association was found between teachers´ use of RC practices and children´s anxiety/fearfulness. Further tests were conducted to examine teacher-child closeness as a mediator of the relation between RC practices and school performance (e.g., reading) revealing no associations.
Regression analyses showed interaction effects between risk and RC practice, suggesting differential sensitivity of “at risk” children to RC practices. In some cases, RC appeared to show protective effects. In other cases, RC practices leveraged off of existing strengths. Discussion will place these findings in the larger context of the ways in which school environments deflect risk.
CONCURRENT 8, EMERGING OPPORTUNITIES FOR PREVENTION RESEARCH, Organized symposia
DEPRESSION AND DOMESTIC VIOLENCE IN HOME VISITATION
Chair: Frank Putnam
DEPRESSION AND DOMESTIC VIOLENCE IN HOME VISITATION. Frank W. Putnam1, David Olds2, 1Children's Hospital Medical Center, Cincinnati, Cincinnati, OH United States; 2University of Colorado Health Sciences Center, Denver, CO United States
Home visitation has emerged as a promising approach to optimizing child development, enhancing maternal life course, and preventing child maltreatment. It has recently been recognized, however, that there are several important impediments to fully benefiting from this prevention approach. In particular, maternal depression and domestic violence have been cited as major challenges for home visitation. Research indicates that the risk for developing depression in low income, young mothers is 26%, which is a twofold increase relative to the general population of postpartum women. Maternal depression has a negative impact on social, emotional, and cognitive development in children. Moreover, it undermines delivery of psychoeducational prevention approaches such as home visitation. Domestic violence is also relatively common in the population typically served by home visitation. Here, too, domestic violence mitigates the benefits of home visitation. This symposium will present recent research findings about the prevalence of maternal depression and domestic violence within the context of home visitation, and consider innovative approaches to ameliorating these problems. Ammerman et al. will present prevalence and course data on depression in Every Child Succeeds (a community-based home visitation program), and present evaluation findings for In-Home Cognitive Behavior Therapy (IH-CBT), a new treatment for depression that is explicitly integrated with ongoing home visitation services. MacMillan will describe a new initiative to incorporate and evaluate screening for domestic violence by nurses in home visitation. Zeanah et al. will present findings from a clinical trial investigating the impact of the Nurse Family Partnership home visitation model on maternal depression and domestic violence. The symposium will be chaired by Frank W. Putnam, M.D., and the Discussant will be David Olds, Ph.D.
IN-HOME CBT FOR DEPRESSED MOTHERS IN HOME VISITATION. Robert T. Ammerman1, Frank W. Putnam1, Jack Stevens2, Judith B. Van Ginkel1, 1Cincinnati Children's Hospital Medical Center, Cincinnati, OH United States; 2Columbus Children's Hospital, Columbus, OH United States
Examination of depression prevalence in 806 mothers enrolled in Every Child Succeeds (a community-based home visitation program using both the Healthy Families America and Nurse Family Partnership models) revealed that 44.2% of mothers had elevated levels of depression (14 or higher on the Beck Depression Inventory-II) during the first year of service. Yet, only 18% of depressed mothers received mental health treatment during that time. In response, we developed and pilot tested In-Home Cognitive Behavior Therapy (IH-CBT), a novel intervention designed specifically for use in the context of home visitation. IH-CBT has four core features: (1) it is based on the empirically-supported CBT model for depression developed by Beck and colleagues, (2) it is administered in the home to overcome barriers to obtaining effective treatment in clinic settings, (3) IH-CBT is designed to address the specific needs and concerns of depressed first-time mothers, and (4) there are systematic procedures to ensure synergistic collaboration between the IH-CBT therapist and home visitor in order to maximize the benefits of both efforts. IH-CBT was evaluated in 26 first time mothers in home visitation who, during the first year of service, scored 20 or higher on the BDI-II and were diagnosed with major depression using the PRIME MD. The sample had high pre-treatment BDI-II scores (mean = 30.35), and exhibited histories of interpersonal trauma (73%), suicide attempts (35%), and hospitalizations (50%). IH-CBT was administered over 17 weekly sessions, two of which were conducted with the home visitor present. Results indicated that 69% of mothers no longer met criteria for major depression at the end of treatment; an additional 15% were partially remitted. A 16 point reduction was found on the BDI-II (F(1, 25) = 42.66, p < .001). Additional findings included decreased self-reported functional impairment, increased self-reported acceptance of and emotional closeness to the child, increased rate of home visiting during treatment period, and overall satisfaction with the program by mothers and home visitors. Results are discussed in terms of the need to further refine this approach to maternal depression, and a proposed clinical trial will be described to subject IH-CBT to a more rigorous design.
ASKING ABOUT AND RESPONDING TO INTIMATE PARTNER VIOLENCE IN HOME VISITATION. Harriet MacMillan1, 1McMaster University, Hamilton, Ontario Canada
Violence against women is prevalent and associated with significant impairment, yet little is known about effective ways to intervene once it is identified. Furthermore, several authors have recommended screening women for violence despite recent guidelines by both the Canadian Task Force on Preventive Health Care and the US Preventive Services Task Force concluding that there is insufficient evidence to recommend for or against screening. As part of the preliminary work leading up to a randomized controlled trial to evaluate whether screening for violence against women in health care settings is effective in reducing violence, we have been assessing performance, feasibility and acceptability of approaches to screening. The randomized trial currently underway will determine which screening instrument (Partner Violence Screen and Woman Abuse Screening Tool) and approach (paper/pencil versus face-to-face versus computer-based) should be used in the forthcoming trial of screening effectiveness. One of the settings in which this trial is being conducted is a program of home visitation provided by nurses post-natally to mothers identified as high risk for parenting problems. Within this context, we have been collecting information about mothers´ perceptions of being asked about IPV and their preference for approach using these screening instruments. Disclosure of exposure to violence is then compared to the “gold standard” instrument: the Composite Abuse Scale. Nurses´ impressions of asking about such exposure and then responding to it if present are being assessed through self-report questionnaires and individual qualitative interviews. In addition, nurses are reporting their perceptions of training they received regarding IPV, as well as degree of comfort in asking. Results from this trial, to be completed by December 2004, will provide new knowledge about the attitudes and preferences of high-risk new mothers regarding being asked about IPV. Information from the nurses will assist us in understanding the extent to which they feel comfortable asking, the methods they prefer and how well prepared they are to respond. Although there are no evidence-based services to which nurses can refer mothers exposed to violence, this issue will arise in the context of home visitation. As rigorous studies to evaluate the effectiveness of approaches to preventing violence are conducted, we need to understand the attitudes of mothers and the nurses who visit them regarding being asked about exposure to violence. Implications of these findings for nurse home visitation will be discussed.
AUGMENTING THE NURSE-FAMILY PARTNERSHIP TO ADDRESS MATERNAL MENTAL HEALTH IN PREGNANCY: RESULTS FROM THE LOUISIANA FIELD TRIAL. Paula D. Zeanah1, Neil W. Boris1, Julie A. Larrieu1, Geoffrey A. Nagle1, 1Tulane Univeristy Health Sciences Center, New Orleans, LA United States
In this presentation, we describe the results of a small randomized-controlled field trial of the Olds et al. Nurse-Family Partnership conducted in Louisiana using updated measures of maternal mental health functioning. Mothers were recruited early in pregnancy (1999-2000) and screened for depression (using the Beck Depression Inventory) and partner violence (using the Partner Violence Inventory). This presentation focuses on prenatal findings. Women exposed to the intervention reported significantly lower depression scores at 28-34 weeks gestation, whether looking at mean scores [t(204) = -2.73, p = .007] or using an established screening cutoff for pregnancy [÷2 (1, n = 206) = 6.086, p = .015, OR = 2.20, 95% CI = 1.17, 4.13]. However, there were no differences in mean partner violence scores between intervention and control groups for women who reported being victimized by a current partner [t(197)= -1.79, p = .076], or perpetrating against a current partner [t(197) = -0.91, p = .363]. While preliminary results here showed a treatment effect on depressive symptoms, more research is needed to determine effective methods of mitigating partner violence. Our experience of augmenting nurse teams in Louisiana by providing a half-time specially-trained mental health provider will be discussed in light of the need for effective programs to prevent maternal depression and impact partner violence. The pregnancy period offers a window of opportunity to engage at-risk mothers and the Nurse-Family Partnership is a well-tested intervention for delivery of interventions targeting depression and partner violence.
CONCURRENT 9, METHODS, Organized symposia
ADVANCED APPLICATIONS OF DISCRETE-TIME SURVIVAL ANALYSIS SYMPOSIUM
Chair: Deborah Drabick
ADVANCED APPLICATIONS OF DISCRETE-TIME SURVIVAL ANALYSIS SYMPOSIUM. Deborah Drabick1, Katherine Masyn2, 1Temple University, Philadelphia, PA United States; 2University of California, Davis, Davis, CA United States
This symposium brings together advanced applications of discrete-time survival analysis in prevention science. Although not a new method, the use of discrete-time survival analysis has increased more notably the last few years due to accessibility of software and methodological dissemination in the applied literature. Further methodological extensions now available allow applied researchers the flexibility to specify more complex event history models that are better able to reflect the intricacies of the event processes under study. For example, placing the discrete-time survival model within a latent variable framework (see Muthén & Masyn, 2004) allows for a much broader class of event history models that utilize the many facets of that more general framework.
The three papers presented in this symposium illustrate some of these newly accessible extensions of discrete-time survival analysis. Specifically, this collection of papers demonstrates the applications of 1) discrete-time survival mixture analysis; 2) multilevel discrete-time survival analysis; and 3) joint discrete-time survival analysis and growth mixture modeling. Each of the papers utilizes a particular extension to answer a substantively salient set of research questions and each present the corresponding analysis results with consideration of the implications for prevention.
DISCRETE-TIME SURVIVAL MIXTURE ANALYSIS OF SUBSTANCE USE INITIATION. Eric Brown1, Charles Fleming1, Richard Catalano1, Katherine Masyn2, 1University of Washington, Seattle, WA United States; 2University of California, Davis, Davis, CA United States
Predictors of marijuana, alcohol, and cigarette use onset were examined using discrete-time survival mixture analyses (DTSMA; Masyn 2003; Masyn & Muthén, 2004). DTSMA allows for examination of unobserved heterogeneity in the distributions of substance use initiation and tests for covariate effects--both within and between latent survival classes--in a fully latent variable framework. Consistent with prior research on substance use initiation, three models were examined: (a) a one-class model assuming a common hazard distribution for all students, (b) a two-class model that partitioned the hazard distribution of substance use onset into two survival classes consisting of long-term abstainers and initiators; and (c) a three-class model consisting of long-term abstainers, early initiators, and late initiators. The sample for these analyses consisted of 1,009 students in the Raising Healthy Children project, a 12-year longitudinal study examining the etiology and consequences of student problem behaviors. Data for this study consisted of students´ annual self-reported use of marijuana, alcohol, and cigarettes. Results indicated that age onset of marijuana and alcohol use were best modeled by multiple survival classes. Specifically, the distributions of initial use of these substances were best represented by two-class models consisting of long-term abstainers and initiators. For initial cigarette use, however, the data were best modeled by a single population distribution (i.e., no mixture distributions). Examination of between-class effects indicated that alcohol and marijuana initiators were significantly more likely to have higher baseline levels of antisocial behavior and family history of alcohol use or cigarette smoking than long-term abstainers. Among alcohol and marijuana initiators, examination of within-class effects indicated that earlier onset of use also was associated with higher baseline levels of child depression. Extensions of this model will incorporate distal outcome variables to examine onset of substance use as a mediator of putative risk and protective factors for substance use related problems and dependency at age 18.
MODELING TIME TO SCHOOL REMOVAL IN PUBLIC SCHOOL STUDENTS. Hanno Petras1, Katherine Masyn2, Sheppard Kellam3, Jacquelyn Buckley4, 1Johns Hopkins University, Baltimore, MD United States; 2University of California, Davis, Davis, CA United States; 3American Institutes of Research and Johns Hopkins University, Washington, DC United States; 4Johns Hopkins Bloomberg School of Public Health, Baltimore, MD United States
Removing students from the educational environment, as a form of punishment, has become one of the most common forms of discipline in American public schools (Dupper, 1994; Dupper & Bosch, 1996; Skiba & Knesting, 2002). In fact, over three million students are suspended each year (Department of Education Office Civil Rights, 2000) despite the lack of clear evidence regarding the effectiveness of school removal in deterring future misbehavior. Furthermore, it is well documented that removed students are more likely to repeat a grade, to prematurely drop out of high school, and are more likely to get involved with the juvenile and adult justice system. Investigations regarding early predictors, which can inform future interventions, are therefore warranted.
For this analysis, 1169 urban public school youth who participated in a larger preventive intervention trial were selected (PI: S. Kellam). Discrete-Time Survival Analysis (DTSA) using latent variables (Muthén & Masyn, 2004), implemented in the Mplus software (Muthén & Muthén, 1990-2004), was used to model the hazard probability of removal between fall of first grade and spring of seventh grade and to determine the predictive power of selected status variables (i.e., gender, race, and poverty). In addition, the extent to which the youth's individual behavioral adaptation to the first grade classroom (i.e., aggressive/disruptive behavior) as well as the overall classroom-level of behavioral problems contributed to the risk for school removal was explored in this modeling framework.
Initial investigation of the effects of gender, poverty and race on the hazard of school removal using a proportional hazard odds model revealed a significant increase in hazard for males who are African-American and who receive subsidized lunch. Extending this model, fall of first grade levels of individual as well as classroom level of aggressive behavior were added to the model. Adjusting for the three status variables, early aggression also increased the hazard for later removal while high levels of classroom aggression appear to reduce the risk for later removal. To better understand the influence of individual and contextual predictors, a more rigorous multilevel DTSA analysis is presented. This approach allowed for the testing of cross-level effects and interactions, e.g., whether the average level of peer aggression in the first grade classroom mediates the effect of individual level aggression in first grade on the time-to-school-removal. The results are discussed regarding the preventive implications for the teacher-student interaction and the later risk for school removal.
DISCRETE-TIME SURVIVAL MIXTURE ANALYSIS OF POST-TREATMENT DRINKING LAPSES AND CONSEQUENT DRINKING TRAJECTORIES WITH IMPLICATIONS FOR RELAPSE PREVENTION. Katie Witkiewitz1, Katherine Masyn2, 1University of Washington, Seattle, WA United States; 2University of California, Davis, Davis, CA United States
Relapse has been a vexing problem in the treatment of addictive behaviors. One of the common assumptions is that relapse (the return to heavy drinking following a period of reduced drinking or abstention) is inevitable following treatment for an alcohol use disorder. The goal of the current study was to examine this assumption and gain a better understanding of the relapse process. Focusing on individual differences in drinking rates it was proposed that unique relapse trajectories can be used to characterize variation in post-treatment drinking. Further, it was hypothesized that the amount of time to the first lapse and the trajectory of drinking following the first lapse could be modeled in a general latent variable framework using several risk factors as predictors of the latent drinking trajectories. Discrete time survival analysis and latent growth mixture modeling were used to model the individual variation in the time-to-lapse and the post-lapse drinking trajectories, respectively. The data for the current study was obtained from the Relapse Replication and Extension Project, which included 563 individuals assessed monthly for the first 12-months following community alcohol treatment. In general, the results from these analyses indicate significant heterogeneity in post-treatment drinking patterns. The findings support a four trajectory model, including heavy, light, heavy-to-light, and moderate drinking trajectories. Contradicting the historical view of relapse, the return to heavy drinking is not the most common outcome. Rather, the current findings demonstrate a large initial lapse followed by a return to abstinence or light drinking is the rule, not the exception. The discrete time survival analysis showed coping, self-efficacy, the abstinence violation effect, and negative affect as significantly related to the time to the first lapse in the expected direction. The 4-class growth mixture model provided the best fit to the actual data for both quantity and frequency of drinking following the first lapse, however the relationships between risk factors and drinking frequency or quantity were unique. Namely, coping was the strongest predictor of drinking frequency trajectories, with the heaviest drinkers reporting the lowest coping scores. Negative affect, distal risks and cognitive factors (self-efficacy, outcomes expectancies, and the abstinence violation effect) were predictive of drinking quantity, with the heaviest drinkers reporting the highest negative affect and distal risks, and the lowest self-efficacy, highest outcome expectancies and higher abstinence violation effects. The implications of these findings for treatment providers, relapse prevention, and future research will be discussed.