Health Education Research, Vol. 18, No. 1, 74-87,
February 2003
© 2003 Oxford University Press
One-year follow-up results of the STARS for Families alcohol prevention program
Center for Drug Prevention Research, University of North Florida, 4567 St Johns Bluff Road South, Jacksonville, FL 32224-2645 and1 Department of Psychology, University of Maryland–Baltimore County, Baltimore, MD 21228-5398, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMethodResultsDiscussionReferences |
|---|
This study examined the 1-year follow-up effects of the STARS (Start Taking Alcohol Risks Seriously) for Families program, a 2-year preventive intervention based on a stage of acquisition model, and consisting of nurse consultations and parent materials. A randomized controlled trial was conducted, with participants receiving either the intervention or a minimal intervention control. Participants included a cohort of 650 sixth-grade students from two urban middle schools—one magnet (bused) and one neighborhood. Trained project staff administered questionnaires to students following a standardized protocol in the schools. For the magnet school sample, significantly fewer intervention students (5%) were planning to drink in the next 6 months than control students (18%),
2 = 11.53, 1 d.f., P = 0.001. Magnet school intervention students also had less intentions to drink in the future, greater motivation to avoid drinking and less total alcohol risk than control students, Ps < 0.05. For the neighborhood school, intervention students (m = 7.90, SD = 1.87) had less total alcohol risk than control students (m = 8.42, SD = 1.83), F(1,205) = 4.09, P = 0.04. These findings suggest that a brief, stage and risk/protective factor tailored program holds promise for reducing risk for alcohol use among urban school youth 1 year after intervention, and has the unique advantage of greater ‘transportability’ over classroom-based prevention programs. |
Introduction |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
A recent study by the US Substance Abuse and Mental Health Services Administration indicated that adolescents who drink alcohol are more likely to experience a range of behavioral problems including delinquent, aggressive and criminal behaviors (Substance Abuse and Mental Health Services Administration, 2000
). In particular, among a sample of youth aged 12–17 years, a strong relationship was found between alcohol consumption and problems such as fighting, stealing, driving after drinking, skipping school, being depressed and deliberately trying to hurt or kill oneself. In addition, alcohol use impacts youth not only directly through use, but also indirectly through environmental exposure to alcoholrelated family violence and community crime (Young, 1997).
The development of health behavior, including alcohol and drug use, has increasingly been described as a multi-stage process (Weinstein et al., 1998; Armitage and Conner, 2000). Stage models are important in that they suggest the tailoring and sequencing of interventions to individuals based on their stage status and mediators influencing stage movement. Tailored interventions hold promise for being more effective and efficient than traditional ‘one size fits all’ generic programs (Skinner et al., 1999).
A variety of theoretical models have been proposed for alcohol, tobacco and other drug use initiation; however, most have lacked some characteristic of true stage theory (Weinstein et al., 1998), such as identifying the ordering of stages and potential factors for influencing stage movement. Furthermore, previous stage models have been hampered by other limitations, including omitting important preconsumption motivational stages and either focusing too narrowly on describing the stages of one drug type alone, e.g. cigarette use, or too broadly on ‘drugs’ in general, without recognition of differences in inter-drug stage acquisition across substance type (Werch and Anzalone, 1995).
One acquisition stage framework that does not suffer from the aforementioned limitations is the Multi-Component Motivational Stages (McMOS) model (Werch and DiClemente, 1994). McMOS has hypothesized stages of habit acquisition of harm-producing behaviors, analogous to the stages of habit change described in the Transtheoretical Model (TM) (Prochaska and DiClemente, 1983). In a recent review of research examining the stages of initiation for alcohol use (Werch, 2001), findings from seven cross-sectional studies suggested that there are youth within each of the stages of alcohol acquisition hypothesized in the McMOS prevention model, and that stage status is significantly associated with a number of risk and protective factors. The majority of studies examining alcohol initiation have used some variation of the stages described in the McMOS model (Westhoff et al., 1996; Migneault et al., 1997; Kelley et al., 1999).
The McMOS prevention model differs from the TM in a number of important ways. First, McMOS proposes a set of stages of acquisition that parallel and exist in conjunction with the stages of change, thereby describing 10 stages—five in acquisition, along with five in change. The two levels of stages reflect different prevention goals (i.e. retarding versus promoting stage movement) and are therefore qualitatively different. Second, while the TM is founded on theories taken from psychotherapy, the McMOS prevention model hypothesizes that progression through the stages of acquisition and change is influenced by risk and protective factors described as constructs within contemporary psychosocial health theories. These theories include, but are not limited to, the Health Belief Model (Becker, 1974), Social Cognitive Theory (Bandura, 1986) and Behavioral Self-Control Theory (Kanfer, 1975). Third, McMOS differs from the TM in that it proposes a range of communication channels for matching prevention content and strategies to specific stage status. In particular, three categories of delivery modes are proposed, including media and media-related materials, interpersonal, and environmental channels.
An underutilized strategy for intervening with youth for alcohol use prevention is the application of primary health care providers outside of typical clinical settings. Several studies have examined the role of the physician and nurse in influencing cigarette smoking (Taylor et al., 1996; Richmond and Mendelsohn, 1998) and alcohol abuse (Burge et al., 1997; Fleming et al., 1997; McIntosh et al., 1997; Watson, 1999). Little research exists, however, to support whether health care providers are effective in preventing the initiation of alcohol use among youth populations (Schonberg, 1988; Simons-Morton et al., 1992). While primary health care providers have been encouraged to take a greater part in advising and educating patients to modify behavioral risk factors, only rarely do they appear to interface with youth in non-clinical settings for the purpose of providing preventive interventions.
A series of pilot studies examining the potential efficacy of behavioral stage-matched preventive interventions involving the use of physician and/or nurse consultations with children (Werch et al., 1996a,b, 1998, 1999) suggest that the use of brief health care provider interventions provide a potentially feasible alternative to more extensive curriculum-based prevention programs. Prior studies have examined the efficacy of the STARS (Start Taking Alcohol Risks Seriously) for Families program at 1- (Werch et al., 2001) and 2- (Werch et al., 2000) year post-test periods of the intervention. In this follow-up study, we hypothesized that students exposed to the intervention would demonstrate less risk for alcohol use than those in the control group 1 year after the conclusion of the intervention.
|
Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Subjects
Participants included a cohort of 650 sixth-grade students from two middle schools located in the economically disadvantaged inner city of Jacksonville, FL. Students were recruited by project staff during the fall of 1996 to participate in a study examining the efficacy of an alcohol prevention program and were followed for 3 years until the end of their eighth grade year in school. A total of 388 students were recruited from a magnet (bused) school and 272 students were recruited from an inner-city neighborhood school. Of all sixth-grade students enrolled, 87% were successfully recruited to participate in the study from each school by returning signed parental consent and student assent forms. Ten students at the neighborhood school did not complete baseline questionnaires, for a total of 650 subjects.
Of the participating students, the majority were African-American (58%), followed by Caucasian (34%) and other racial classifications (8%). Over half of the students (54%) were male and the mean age of all students was 11.40 years old (SD = 0.71). Over half (55%) were in the free lunch program, indicating that their families were economically disadvantaged. Nearly one-third of students (31%) reported perceiving an alcohol or drug problem in the immediate family and over four in 10 (42%) reported not having any alcohol or drug education during the last year prior to intervention implementation.
Procedures
Students were randomly assigned within each school to either the intervention (STARS for Families program) or a minimal intervention control. Baseline (beginning fall semester 1996) and annual post-test data (concluding spring semesters 1997 and 1998) were collected at the school sites during the 2-year intervention, the results of which have been published elsewhere (Werch et al., 2000, 2001). One-year follow-up data (concluding spring semester 1999) were also collected at the target schools and serve as the focus of this study.
Students returning signed informed consents and assents were asked to complete a confidential youth questionnaire. Immediately prior to the administration of the questionnaire, a ‘dipstick’ saliva pipeline procedure (Alco Screen; Chematics, Inc.) was used to increase the validity of self-reported alcohol use (Pechacek et al., 1984; Botvin et al., 1990; Johnson et al., 1990). Trained project staff administered questionnaires to students at the targeted schools following a standardized protocol emphasizing confidentiality of participant reports. Follow-up data were collected using procedures and measures identical to those during the baseline data collection. This research protocol was approved by the University of North Florida’s institutional review board prior to implementing the study.
Intervention
The intervention examined in this study was founded on the McMOS prevention model (Werch and DiClemente, 1994). Students assigned to receive the STARS for Families program were provided with a 2-year, multicomponent intervention. Prevention messages addressed specific stage status and risk/protective factors of individual youth, based on preintervention data collected using the Youth Alcohol and Drug Survey (Werch, 1996).
During the fall semester of the sixth grade, intervention youth received a brief one-on-one health consultation provided by a nurse about why and how the child should avoid alcohol use. During the spring semester of the sixth grade, intervention youth received a series of prevention postcards mailed to parents/guardians providing key facts on what to say to their children about avoiding alcohol. During the fall semester of the seventh grade, intervention youth received a follow-up nurse consultation. Lastly, during the spring semester of the seventh grade, intervention youth received four family take-home lessons providing activities to enhance parent–child communication regarding prevention skills and knowledge.
Trained nurses provided health consultations using standardized protocols. Each protocol included directions for implementing the consultation, an RN interview form providing stage-matched prevention messages, title of the risk/protective factor, the purpose of the risk communication, prevention messages addressing that risk/protective factor and a nurse recommendation contract asking the child to avoid future alcohol use. The consultation protocols used a checklist format, designed to better ensure that all of the prevention content was reviewed with the client. As many as 12 risk and protective factors were addressed during the health consultation, based on annual preintervention stage status and risk/protective factor data. These included perceived susceptibility and severity; perceived benefits; cues to avoid alcohol; motivation and intentions; the environment and influenceability; self-efficacy and behavioral capability; situation (perceived prevalence); expectations; expectancies; emotional coping responses; observational learning; and self-evaluation, self-monitoring and self-reinforcement. For example, students in a preparation, action or maintenance stage of initiating alcohol use were provided with a prevention message addressing emotional coping responses to deal with stress that could lead to alcohol use.
Nurses received a 1-day training that included demonstrations, role playing and feedback from project staff on how to implement the consultations. Nurse consultations took approximately 20 min to implement. Second-year follow-up nurse consultations used the same standardized protocol employed in implementing the first year of the intervention, thereby providing a booster session.
Physician-endorsed mailed prevention postcards were addressed to the parent or guardian of participating youth. Postcards requested that the parent/guardian take a few minutes to read and talk about the important key fact found on the card, to help the child continue to stay away from alcohol. Each trifold postcard was color coded to identify a new key fact, which addressed a particular risk/protective factor.
Parents/guardians were mailed up to 10 postcards, based on youth stage status and the number of risk/protective factors identified during annual preintervention data collections. The risk/protective factors addressed were the same as in the health consultations, and included perceived benefits, cues to avoid alcohol, and motivation and intentions; perceived susceptibility and severity; environment and influenceability; situation (perceived prevalence); expectations; expectancies; self-efficacy and behavioral capability; self-evaluation, self-monitoring and self-reinforcement; observational learning; and emotional coping responses. For example, the key fact message for the risk factor of expectations was worded ‘Tell your child that you would be very upset if he or she drank alcohol. Research shows that most kids do NOT like their friends to drink alcohol either!’. Two postcards were mailed per week, beginning in January of the spring semester. Each card was endorsed and signed by a local pediatrician and the principal investigator.
Physician-endorsed family-based lessons provided a set of brief activities for parents/guardians and children to complete together. Each of the four modular lessons included a cover letter describing the program, a set of three activity sheets, a contract that asked the child to make a promise to avoid alcohol use, and a feedback sheet to collect process data regarding parent/guardian use of and satisfaction with each lesson. Each of the four lessons was color coded to identify a new module. Parents/guardians and children were asked to complete each of the lessons together. In addition, they were informed that by returning completed lessons they could earn chances to win prizes. The cover letter was endorsed and signed by a local pediatrician and the principal investigator.
Two lessons addressed risk factors for alcohol use (targeted for reduction) and two lessons addressed protective factors (targeted for enhancement), paralleling those factors addressed during each of the earlier intervention components. The two risk-factor lessons covered environment and influenceability, situation (perceived prevalence), and expectancies. The two protective-factors lessons covered perceived susceptibility and severity, expectations, and self-efficacy and behavioral capability. Activities found within the modules for both parents/guardians and children to complete included true/false statements, check-off lists, fill in the blank items, listing activities and role-playing activities. The contract found within each lesson asked the child to make a promise to stay away from alcohol each day during the next week and provided space for the name of the parent/guardian who would remind the child of this important pledge. One lesson was administered per week for 4 consecutive weeks.
Minimal intervention control
Students assigned to the minimal intervention control condition were given alcohol education booklets including Young People and Alcohol—What the Ads Don’t Tell You during the fall semester of the sixth grade intervention and The Truth About Alcohol (Channing L. Bete Co., Inc.) during the fall semester of the seventh grade intervention. These booklets included information concerning alcohol’s effects on the body, risks of using alcohol for youth, reasons why youth drink, reasons not to drink alcohol, ways of refusing alcohol use offers, alternatives to drinking, learning to feel good about oneself, the stages of intoxication, types of drinkers, the characteristics of alcohol abuse, the effect of alcohol on health, and other questions and answers about alcohol. Control students were given a booklet, placed in a quiet area during the health consultations, and asked to read the material on their own. The average time for students to complete reading the control materials was approximately 10 min.
Measures
The 77-item Youth Alcohol and Drug Survey (Werch, 1996) was used to collect data on alcohol and drug consumption, and alcohol-related cognitive, social and behavioral risk and protective factors. The questionnaire took approximately 20 min to complete. Extensive pilot testing of the questionnaire resulted in a highly readable, appropriate and efficient instrument for the target population, as well as a psychometrically sound instrument with acceptable estimates of validity and reliability. Test–re-test reliability of the instrument was 0.99, using a sample of 22 sixth- to eighth-grade students who were administered the instrument twice with 2 weeks between administrations. Alcohol consumption patterns were measured from items adopted from previous alcohol abuse prevention research (Botvin et al., 1984; Johnson et al., 1990; Ellickson and Hays, 1991) and for this analysis included items measuring lifetime use (ever used); drinking during the last year; how long one has been drinking alcohol; 30- and 7-day frequency of use; 30- and 7-day quantity of use; heavy drinking, defined as consuming 5 or more drinks in a row during the last 30 days and 2 weeks; nine items measuring negative consequences experienced during drinking; and four items measuring intentions to think about, plan, try and use alcohol in the next year. Alpha coefficients for these measures included 0.80 for the frequency of alcohol use items, 0.80 for the quantity of alcohol use items, 0.95 for the heavy drinking items, 0.88 for the alcohol consequences items and 0.75 for the intentions items. Items measuring similar constructs were summed to create combined measures of current frequency, quantity, heavy alcohol use, alcohol consequences and intentions to drink in the future.
Alcohol use initiation during the last year was measured using an item that asked, ‘During the last year, did you start drinking alcohol?’. This item, adopted from previous stage research and theory (Prochaska and DiClemente, 1992; Werch and DiClemente, 1994; Werch and Anzalone, 1995), had five response categories, reflecting the stages of initiation including (a) I did not try it last year (precontemplation), (b) I am thinking of trying alcohol soon (contemplation), (c) I am planning to start drinking soon (preparation), (d) I started drinking during the last 6 months (action) and (e) I have been drinking for longer than 6 months (maintenance).
Social, behavioral, environmental and cognitive risk and protective factors associated with the three behavioral theories underpinning the McMOS prevention model (Werch and DiClemente, 1994) were also measured. Because of the rather large number of risk factor measures, only those most highly correlated with pre-test alcohol use measures were selected as dependent variables. These included measures of motivation to avoid drinking, expectancy beliefs, peer prevalence, influenceability and total risk factors for alcohol use.
Two items were used to measure motivation to avoid alcohol during the next 30 days and year. These items had an 
coefficient of 0.92. Eleven outcome-expectancy beliefs were measured, asking what outcomes subjects thought using alcohol resulted in. These items had an coefficient of 0.76. Two items were used to measure perceived peer prevalence of drinking. These items asked how many friends, and how many kids at their school, drank alcohol. These items had an coefficient of 0.47. Three items were used to measure influenceability, with an coefficient of 0.82. In addition, a total alcohol risk factors measure was formed by first combining similar measures, then setting cut-off scores to determine risk or no-risk categories (Werch et al., 1997), and finally adding the total number of risk categories. Specifically, measures which were combined included perceived susceptibility and severity; motivation to avoid and intentions to drink; environment (consisting of opportunities to drink, family history of alcohol or drug problems and siblings who drink alcohol) and influenceability; resistance self-efficacy and behavioral capability (refusing alcohol offers); peer and parent expectations; and situation (perceived peer and adult prevalence of drinking). In addition, individual measures of self-control practices, perceived benefits and expectancies were included as risk factors. Next, risk-factor status categories were determined for each measure by selecting liberal cut-off scores to identify those students demonstrating any risk potential. For example, the expectancy risk factor cut-off score was set to identify all students who responded that alcohol resulted in any positive consequence. Lastly, a total alcohol risk factors score was constructed by adding across all nine combined risk factor categories for each subject, with total alcohol risk factors ranging from 0 (least risk) to 9 (greatest risk).
Various sociodemographic measures were also collected. These included ethnicity, gender, age, participation in a free lunch program, exposure to previous alcohol or drug education programs, parents or guardians the youth lived with and perceived alcohol or drug problem in the immediate family.
Data analysis strategy
Data were analyzed using statistical procedures contained in SPSS for Windows 95, Release 10.0. Summary statistics including means, standard deviations and frequencies were used to describe the data. Because of numerous subject differences between the two schools, school-site data were analyzed as separate samples. Pre-test and selected follow-up alcohol/drug use and risk/protective factor data were analyzed using 2 analyses for dichotomous variables and ANOVAs for continuous measures. Follow-up outcome data were analyzed using MANOVAs, first for alcohol use measures and then for alcohol use risk/protective factors. MANOVAs were used to control for type I errors over the two dependent variable sets. Since few significant differences were found between intervention groups at baseline, MANCOVAs were generally deemed unnecessary to adjust for pre-test differences. Although significant differences between schools on baseline socio-demographic and alcohol use measures indicated separate samples, MANOVA and MANCOVA analyses were also conducted examining collapsed school data due to small sample sizes for individual schools. Lastly, repeated measures MANOVAs were run primarily as a confirmatory analysis to the 1-year follow-up MANOVA analyses by examining the smaller sample of subjects for which data were collected at all data points. Post hoc t-tests were then conducted for significant interaction effects to determine at which data collection points differences were found between intervention and control groups.
|
Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Comparison of school sites
Significant differences were found between students at the two schools on six of the seven demographic measures and three of the five alcohol use measures. Students in the neighborhood school sample were more likely to be female (
2 = 12.08, 1 d.f., P = 0.0005), African-American (2 = 120.10, 6 d.f., P = 0.00001), older (t-test = 7.39, 423.32 d.f., P = 0.001), receive free school lunch (2 = 111.15, 1 d.f., P = 0.00001), live in a single parent/guardian home (2 = 34.61, 4 d.f., P = 0.00001) and receive no prior alcohol or drug education during the past year (2 = 47.19, 1 d.f., P = 0.00001) than were students in the magnet school. In addition, neighborhood students were less likely to have experienced lifetime alcohol use (2 = 7.64, 1 d.f., P = 0.006), greater levels of negative alcohol consequences (t-test = 2.81, 501.72 d.f., P = 0.005) and greater total risk for drinking (t-test = 4.51, 647 d.f., P = 0.001). No differences were found between schools on perceived alcohol or drug problem in the immediate family, alcohol use during the past year and heavy alcohol consumption. Because of the extent of differences between youth at the two schools, each school was treated as a separate sample during initial outcome analyses.
Baseline and attrition analyses
At pre-test, no significant differences were found between intervention and control groups on any of the demographic measures, both within and across schools. Similarly, no significant differences were found between experimental groups on any of the alcohol use measures at baseline, both within and across schools.
Attrition analysis showed that at 1-year follow-up, 78% of the sample completed a questionnaire (n = 507), with a comparable proportion of dropouts occurring from the neighborhood (21%) and the magnet school (23%). Dropouts were evenly distributed between intervention (n = 75) and control (n = 68) groups. In addition, the proportion of dropouts occurring from the intervention group at the neighborhood school (56%) and the magnet school (50%) was similar.
Few differences were found between dropouts and those completing the follow-up questionnaire at baseline on demographic and alcohol consumption measures. For subjects in the magnet school, dropouts were more likely to have been older (m = 11.41, SD = 0.77) than non-dropouts (m = 11.18, SD = 0.47), F(1,384) = 11.70. P = 0.001. In addition, dropouts were more likely to have experienced negative alcohol consequences (m = 0.64, SD = 2.00) than non-dropouts (m = 0.18, SD = 0.94), F(1,384) = 9.05, P = 0.003. No differences were found between dropouts and non-dropouts on demographic and alcohol use measures for neighborhood school students. However, one group-by-dropout status interaction effect was found among students in the magnet school, with intervention students who dropped out having greater intentions to drink (m = 5.36, SD = 2.72) than control students who dropped out (m = 4.77, SD = 1.60), F(1,384), P = 0.04. No interaction effects between group and dropout status were found on any baseline demographic or alcohol use measure for subjects at the neighborhood school.
Process measures
A retrospective process evaluation examined the intervention integrity of the STARS for Families program (Watts et al., under review). Participating students and nurses completed process data collection instruments that asked about the integrity of the program and their satisfaction with its services. Students’ feedback on seven process measures of satisfaction and efficacy indicated that between 93 and 98% of intervention students perceived the nurse consultation favorably on the measures. In contrast, between 83 and 98% of control students perceived the control materials favorably on the same measures. In addition, nurses’ feedback on three measures of efficacy and satisfaction indicated that between 87 and 97% perceived their consultations favorably on the measures. Lastly, an analysis of sampled taped nurse consultations by research staff found that between 80 and 100% of consultations were rated favorably on measures of accuracy, effectiveness, enthusiasm, responsiveness and smoothness.
A pilot study of the efficacy of the parent postcards for increasing parent–child communication about alcohol prevention was recently published (Carlson et al., 2000). Parents were asked to respond to a 10-question telephone survey 8 weeks after the implementation of the postcards. This study found that parents receiving the postcards were more likely than those in the control group to have talked with their child about avoiding alcohol 10 or more times in the past year, 2 = 10.49, 4 d.f., P = 0.03. Intervention parents were also more likely to have talked to their child about avoiding alcohol in the last 30 days than control parents, 2 = 4 d.f., P = 0.01.
Process data associated with the take-home lessons showed that between 94 and 98% of parents talked with their child about individual take-home lessons. Between 95 and 99% of parents said they would suggest other parents get various take-home lessons. Finally, between 92 and 97% of parents signed contracts attached to individual take-home lessons which students returned to school.
Follow-up group comparisons
Subjects’ alcohol use measures at 1-year follow-up by school and group is shown in Table I. For the magnet school sample, significantly fewer intervention students (5%) were planning to drink in the next 6 months than control students (18%), 2 = 11.53, 1 d.f., P = 0.001. In addition, fewer intervention students (13%) were in more advanced stages of alcohol acquisition (i.e. contemplation–maintenance) than were control students (21%) and fewer intervention students (11%) drank alcohol for any length of time (i.e. 30 days to 6 months or more) than did control students (21%), although these differences only approached significance (Ps = 0.06). The other measures of life-time alcohol use, 30- and 7-day use, and 30-day heavy use were not significant, but showed fewer intervention students using than control students. Likewise, within the neighborhood school sample, fewer intervention students used less alcohol than control students for all seven alcohol measures, however, these differences were not significant.
|
Mean alcohol use and risk factor measures at 1-year follow-up by school and group are shown in Table II
. For the magnet school, the overall MANOVA for alcohol use was significant, F(5,294) = 2.82, P = 0.01, as well as the univariate test for intentions, showing intervention students with significantly less intentions to drink in the future (m = 5.56, SD = 2.75) than control students (m = 6.70, SD = 3.77), F(1,298) = 8.95, P = 0.003. Along the same vein, the overall MANOVA for alcohol use risk factors was significant for magnet school youth, F(5,294) = 2.68, P = 0.02. Univariate tests showed intervention students had significantly greater motivation to avoid drinking (m = 2.49, SD = 1.17) than control students (m = 3.00, SD = 1.79), F(1,298) = 8.41, P = 0.004, and less total alcohol risk factors (m = 7.73, SD = 1.83) than control students (m = 8.26, SD = 1.96), F(1,298) = 5.78, P = 0.01. For the neighborhood school, the overall MANOVA for alcohol use risk factors was significant, F(5,201) = 2.18, P = 0.05, as was the univariate test for total alcohol risk factors, with intervention students having less alcohol risk (m = 7.90, SD = 1.87) than control students (m = 8.42, SD = 1.83), F(1,205) = 4.09, P = 0.04. While mean alcohol consumption on all four measures of use was lower for neighborhood students receiving the intervention as compared to control students, these differences were not significant.
|
MANOVAs, as well as MANCOVAs with baseline measures used as covariates, were also run with schools collapsed. These analyses found nearly identical results as those for the magnet school, with univariate tests showing intervention subjects with significantly less intentions to drink in the future, greater motivation to avoid alcohol use and less total alcohol risk factors than control subjects (Ps < 0.05).
In addition to the 1-year follow-up data analyses just described, similar results were found when repeated measures MANOVAs were run with a more limited sample of those who provided data at baseline, two annual post-tests and the 1-year follow-up. Mean alcohol use and risk factor measures over time by group and school are shown in Table III. For the magnet school, the overall MANOVAs for alcohol use main effects for group, F(5, 277) = 2.24, P = 0.05 and time, F(15, 267) = 3.96, P = 0.0001, were significant, but group by time interaction was not significant. A significant interaction univariate test for intentions was found, with control students having significantly greater intentions to drink over time compared to intervention students, F(3,843) = 2.74, P = 0.04. Post hoc t-tests showed significantly less intention to drink among intervention students than control students 3 months after program implementation (seventh grade), t-test = -2.70, 295.96 d.f., P = 0.007, and at 1-year follow-up (eighth grade), t-test = -2.99, 272.65 d.f., P = 0.003. For the neighborhood school, the overall MANOVA for alcohol use main effects for time was significant, F(15,160) = 2.12, P = 0.01, but no differences were found on main effects for group or interaction, nor for univariate interaction effects.
|
For the magnet school, the overall MANOVA for alcohol use risk factors main effects for group was significant, F(5,276) = 2.22, P = 0.05, but time main effects and interaction were not significant. Univariate tests showed interactions of group and time for motivation to avoid alcohol, F(3,840) = 3.19, P = 0.02, and total alcohol risk factors, F(3,840) = 3.75, P = 0.01, with control students having significantly less motivation to avoid and more total alcohol risk over time compared to intervention students. Post hoc t-tests showed significantly greater motivation to avoid alcohol among intervention students than control students 3 months after program implementation (seventh grade), t-test = -2.89, 271.62 d.f., P = 0.004, and at 1-year follow-up (eighth grade), t-test = -2.90, 256.39 d.f., P = 0.004. The t-tests also showed significantly less total alcohol risk among intervention students than control students at 3 months post-program, t-test = -2.07, 315 d.f., P = 0.03, and at 1-year follow-up, t-test = -2.40, 298 d.f., P = 0.01. For the neighborhood school, the overall MANOVA for alcohol use risk factors main effects for time was significant, F(15,159) = 10.18, P = 0.0001, but no differences were found on main effects for group or for interaction effects, as well as no univariate interaction effects.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
These results suggest that a brief preventive intervention tailored to youth stage of alcohol use acquisition, and consisting of an annual one-on-one nurse consultation and family print materials, successfully reduced risk for alcohol use 1 year after the intervention. Specifically, for the magnet school sample, students receiving the intervention showed over one-third less risk for planning to drink than control students, and significantly less intentions to drink in the future, greater motivation to avoid alcohol use and less total alcohol risk factors than control students. For the neighborhood school sample, intervention students showed significantly less total risk factors for alcohol use than control students.
Results from this analysis were similar to those found at the 3-month post-test reported earlier (Werch et al., 2000), with two exceptions. First, within the magnet school, positive intervention outcomes on heavy alcohol use and alcohol use expectancy beliefs found at post-test were not seen 1 year later. These differences may indicate a weakening of the intervention’s effects over time. However, they may also reflect attrition influences. This rationale is supported by data showing intervention effects on total alcohol risk factors actually increased during the 1-year follow-up period. In addition, positive outcomes on intentions and motivations to avoid alcohol use found at post-test were also found at 1-year follow-up. Second, within the neighborhood school, intervention effects on reducing total alcohol risk factors 1 year after the program were not seen earlier at post-test. Thus it appears that the intervention effects increased over time for this sample. Given the majority of youth in the neighborhood school were African-American, this finding appears to support other studies (Wallace et al., 1999; Kann et al., 2000; Parker et al., 2000) which suggest that African-American youth develop alcohol use risk at a slower rate than white and Hispanic youth, thereby taking longer to differentiate intervention effects from controls.
While all alcohol use measures reported at 1-year follow-up were lower for intervention students than control students, most were not significantly different. Given this was a very brief intervention, it may be necessary to repeat it annually to obtain maximum effects. It may also suggest the need to intensify the intervention by increasing the number of nurse contacts or parent participation. Another possibility is that this intervention was somewhat weakened by attempting to tailor content to too many risk and protective factors. Some of our earlier studies of preventive interventions tailored to stage of initiation, but not to risk and protective factors, showed more powerful alcohol consumption outcomes (Werch et al., 1996a,b, 2000). Similarly, others have found that good fitting non-tailored materials can perform as well or better than tailored materials (Kreuter et al., 2000). Another reason for not finding more significant program effects was the relatively small sample sizes in each school, indicating the necessity to increase sample size in future studies to increase power to detect moderate intervention effects. This was particularly the case for the neighborhood school, which was the smaller of the two participating schools.
Another possible reason for the apparent difference in program outcomes by school setting is due to contamination. The weaker effects were found within the neighborhood school where contamination of program effects would have been greatest. The stronger effects found in the magnet school may have resulted, in part, from less within-school contamination. A lesser amount of contamination would result because a significant proportion of the student body (more than 40%) was bussed to school from surrounding suburbs, and were therefore not exposed to after school and social events where drinking might occur.
One limitation of this research was the use of only two participating schools, thereby limiting the generalizability of the study’s findings. A second limitation was that since a group-bydropout status interaction effect was found on intentions to drink among magnet school students, outcome effects on intentions may be questioned. This is unlikely to have been the case, however, for the positive outcomes on the other measures of total alcohol risk and motivation to avoid alcohol use, and the pattern of alcohol consumption and risk/protective factor effects favoring youth exposed to the intervention. A third limitation is that while intention can be an excellent proxy measure for behavior, it is not alcohol consumption itself. As a whole, the program’s effects appear to be largely on modifying risk and indicators of future use of alcohol consumption. If this is the case, future research should examine the effects of this program beyond a 1-year period when one might expect to see behavioral manifestations of earlier intentional and risk measures. Lastly, a power analysis was conducted for both 2 and ANOVA tests which indicated that the study suffered from a lack of power to detect small effect sizes commonly found in preventive intervention studies, even with both school samples combined. This may explain why results of analyses conducted on the collapsed school samples resembled those found in the larger magnet school.
In conclusion, the primary implication of these findings is that a brief, stage and risk/protective factor tailored program holds promise for reducing alcohol use risk among urban school youth 1 year after intervention. However, at this time it is recommended that the program be repeated annually to ensure maximum effects. Another option worthy of examination is the use of this brief intervention in combination with other existing science-based programs to enhance their effects, including community-based programs like Project Northland (Perry et al., 2000), or school-based programs like Life Skills Training (Botvin et al., 2000) or Project AAPT (Adolescent Alcohol Prevention Trial) (Donaldson et al., 1995). This research adds to a number of other epidemiological and intervention studies (Werch, 2001) suggesting the potential utility of the McMOS model for understanding and influencing substance use behavior of youth.
The STARS for Families program uniquely employs primary care nurses to interface with youth in a non-clinical setting to advise and educate students regarding risk and protective factors associated with the initiation of alcohol use. Furthermore, the use of parent- and family-based print materials permits an inexpensive approach to increase parent–child communication regarding prevention issues (Carlson et al., 2000). Combined, this type of a multi-component, yet brief intervention may serve as an alternative to the more common school-based curriculum approach. By using nurses and parent materials, the STARS for Families program may be plausibly disseminated on a large scale, given that it reduces the demand on teachers to implement extensive and complicated prevention curricula during a period of increasing emphasis on school accountability. Therefore, even if the STARS for Families program is found to be somewhat less efficacious than other science-based prevention programs, it has the unique advantage of greater ‘transportability’ (Bauman et al., 2001) over other successful prevention programs. This characteristic of the STARS for Families program makes it more likely to be disseminated within school, clinical and community settings than more time- and training-intensive classroom-based curriculum approaches.
|
Acknowledgments |
|---|
This research was supported by a grant from the National Institute on Alcohol Abuse and Alcoholism (grant AA9283). Thanks to Dr Michael Dunn and Lisa Provencher for their helpful comments on an earlier draft of this manuscript.
|
References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Armitage, C. and Conner, M. (2000) Social cognition models and health behaviour: a structured review. Psychology and Health, 15, 173–189.
Bandura, A. (1986) Social Foundations of Thought and Action: A Social Cognitive Theory. Prentice-Hall, Englewood Cliffs, NJ.
Bauman, K., Ennett, S., Foshee, V., Pemberton, M., King, T. and Koch, G. (2001) Influence of a family-directed program on adolescent cigarette and alcohol consumption. Prevention Science, 1, 227–237.
Becker, M. (1974) The Health Belief Model and Personal Health Behavior. C. B. Slack, Thorofare, NJ.
Botvin, G., Baker, E., Botvin, E., Filazzola, A. and Millman, R. (1984) Prevention of alcohol misuse through the development of personal and social competence: a pilot study. Journal of Studies on Alcohol, 45, 550–552.[ISI][Medline]
Botvin, G., Baker, E., Dusenbury, L., Tortu, S. and Botvin, E. (1990) Preventing adolescent drug abuse through a multi-modal cognitive-behavioral approach: results of a 3-year study. Journal of Consulting and Clinical Psychology, 58, 467–446.
Botvin, G., Griffin, K., Diaz, T., Scheier, L., Williams, C. and Epstein, J. (2000) Preventing illicit drug use in adolescents: long-term follow-up data from a randomized control trial of a school population. Addictive Behaviors, 25, 769–774.[CrossRef][ISI][Medline]
Burge, S., Amodei, N., Elkin, B. and Catala, S. (1997) An evaluation of two primary care interventions for alcohol abuse among Mexican-American patients. Addiction, 92, 1705–1716.[CrossRef][ISI][Medline]
Carlson, J., Moore, M., Werch, C. and Pappas, D. (2000) A pilot intervention to increase parent–child communication about alcohol avoidance. Journal of Alcohol and Drug Education, 45, 59–70.
Donaldson, S., Graham, J., Piccinin, A. and Hansen, W. (1995) Resistance-skills training and onset of alcohol use: evidence for beneficial and potentially harmful effects in public schools and in private Catholic schools. Health Psychology, 14, 291–300.[CrossRef][ISI][Medline]
Ellickson, P.L. and Hays, R.D. (1991). Beliefs about resistance self-efficacy and drug prevalence: do they really affect drug use? International Journal of the Addictions, 25, 1353–1378.
Fleming, M., Barry, K., Manwell, L., Johnson, K. and London, R. (1997) Brief physician advice for problem alcohol drinkers: a randomized controlled trial in community-based primary care practices. Journal of the American Medical Association, 277, 1039–1045.[Abstract]
Johnson, C., Pentz, M., Weber, M., Dwyer, J., Baer, N., MacKinnon, D., Hansen, W. and Flay, B. (1990) Relative effectiveness of comprehensive community programs for drug abuse prevention with high risk and low risk adolescents. Journal of Consulting and Clinical Psychology, 58, 447–456.[CrossRef][ISI][Medline]
Kanfer, F. (1975) Self-management methods. In Kanfer, F. and Goldstein, A. (eds), Helping People Change: A Textbook of Methods. Pergamon Press, New York, pp. 309–355.
Kann, L., Kinchen, S., Williams, B., Ross, J., Lowry, R., Grunbaum, J. and Kolbe, L. (2000) Youth risk behavior surveillance—United States, 1999. Morbidity and Mortality Weekly Report, 49 (SS-5), 1–94.
Kelley, R., Denny, G. and Young, M. (1999) Modified stages of acquisition of gateway drug use: a primary prevention application of the stages of change model. Journal of Drug Education, 29, 189–203.[CrossRef][ISI][Medline]
Kreuter, M., Oswald, D., Bull, F. and Clark, E. (2000) Are tailored health education materials always more effective than non-tailored materials? Health Education Research, 15, 305–315.
McIntosh, M., Leigh, G., Baldwin, N. and Marmulak, J. (1997) Reducing alcohol consumption. Canadian Family Physician, 43, 1959–1967.[ISI][Medline]
Migneault, J., Pallonen, U. and Velicer, W. (1997) Decisional balance and stage of change for adolescent drinking. Addictive Behaviors, 22, 339–351.[CrossRef][ISI][Medline]
Parker, K., Calhoun, T. and Weaver, G. (2000) Variables associated with adolescent alcohol use: a multiethnic comparison. Journal of Social Psychology, 140, 51–62.
Pechacek, T., Fox, B., Murray, D. and Leupker, R.(1984) Review of techniques for measurement of smoking behavior. In Matarazzo, J., Herd, J., Miller, N. and Weiss, S. (eds), Behavioral Health: A Handbook of Health Enhancement and Disease Prevention. Wiley, New York, pp. 729–754.
Perry, C., Williams, C., Komro, K., Veblen-Mortensen, S., Forster, J., Bernstein-Lachter, R., Pratt, L., Dudovitz, B., Munson, K., Farbakhsh, K., Finnegan, J. and McGovern, P. (2000) Project Northland high school interventions: community action to reduce adolescent alcohol use. Health Education and Behavior, 27, 29–49.
Prochaska, J. and DiClemente, C. (1983) Stages and processes of self-change of smoking: toward an integrative model of change. Journal of Consulting and Clinical Psychology, 51, 390–395.[CrossRef][ISI][Medline]
Prochaska J. and DiClemente, C. (1992) Stages of change in the modification of problem behaviors. Progress in Behavior Modification, 28, 184–218.
Richmond, R. and Mendelsohn, C. (1998) Physician’s views of programs incorporating stages of change to reduce smoking and excessive alcohol consumption. American Journal of Health Promotion, 12, 254–257.[ISI][Medline]
Schonberg, S. (1988) Substance Abuse: A Guide for Health Professionals. American Academy of Pediatrics, Elk Grove Village, IL.
Skinner, C., Campbell, M., Rimer, B., Curry, S. and Prochaska, J. (1999) How effective is tailored print communication? Annals of Behavioral Medicine, 21, 290–298.[ISI][Medline]
Simons-Morton, B., Mullen, P., Mains, D. and Tabak, E. (1992) Characteristics of controlled studies of patient education and counseling for preventative health behaviors. Patient Education and Counseling, 19, 175–204.[CrossRef][ISI][Medline]
Substance Abuse and Mental Health Services Administration (2000) Patterns of Alcohol use among Adolescents and Associations with Emotional and Behavioral Problems. Office of Applied Studies, Rockville, MD
Taylor, C., Miller, N., Herman, S., Smith, P., Fisher, L., DeBusk, R. and Sobel, D. (1996) A nurse-managed smoking cessation program for hospitalized smokers. American Journal of Public Health, 86, 1557–1560.
Wallace, J., Forman, T., Guthrie, B., Bachman, J., O’Malley, P. and Johnston, L. (1999) The epidemiology of alcohol, tobacco and other drug use among Black youth. Journal of Studies Alcohol, November, 800–809.
Watson, H. (1999) A study of minimal interventions for problem drinkers in acute care settings. International Journal of Nursing Studies, 36, 425–34.[CrossRef][ISI][Medline]
Weinstein, N., Rothman, A. and Sutton, S. (1998) Stage theories of health behavior: conceptual and methodological issues. Health Psychology, 17, 290–299.[CrossRef][ISI][Medline]
Werch, C. (1996) The Youth Alcohol and Drug Survey. University of North Florida, Center for Drug Prevention and Health Promotion, Jacksonville, FL.
Werch, C. (2001) Preventive alcohol interventions based on a stages of acquisition model. American Journal of Health Behavior, 25, 206–216.[ISI][Medline]
Werch, C. and DiClemente, C. (1994) A multi-component stage model for matching drug prevention strategies and messages to youth stage of use. Health Education Research, 9, 37–46.
Werch, C. and Anzalone, D. (1995) Stage theory and research on tobacco, alcohol, and other drug use. Journal of Drug Education, 25, 81–98.[ISI][Medline]
Werch, C., Anzalone, D., Brokiewicz, L., Felker, J., Carlson, J. and Castellon-Vogel, E. (1996a) An intervention for preventing alcohol use among inner-city middle school students. Archives of Family Medicine, 5, 146–152.[Abstract]
Werch, C., Carlson, J., Pappas, D. and DiClemente, C. (1996b) Brief nurse consultations for preventing alcohol use among urban school youth. Journal of School Health, 66, 335–338.
Werch, C., Carlson, J., Pappas, D., Dunn, M. and Williams, T. (1997) Risk factors related to urban youth stage of alcohol initiation. American Journal of Health Behavior, 21, 377–387.
Werch, C., Pappas, D., Carlson, J. and DiClemente, C. (1998) Short- and long-term effects of a pilot program to reduce alcohol consumption. Substance Use and Misuse, 33, 2303–2321.
Werch, C., Pappas, D., Carlson, J. and DiClemente, C. (1999) Six-month outcomes of an alcohol prevention program for inner-city youth. American Journal of Health Promotion, 13, 237–240.[ISI][Medline]
Werch, C., Carlson, J., Pappas, D., DiClemente, C. and Carbonari, J. (2000) Evaluation of a brief alcohol prevention program for urban school youth. American Journal of Health Behavior, 24, 120–131.
Werch, C., Carlson, J., Owen, D., DiClemente, C. and Carbonari, J. (2001) Effects of a stage-based alcohol preventive intervention for inner-city youth. Journal of Drug Education, 31, 123–138.[CrossRef][ISI][Medline]
Westhoff, W., McDermott, R. and Harokopos, V. (1996) Acquisition of high-risk behavior by African-American, Latino, and Caucasian middle-school students. Psychological Reports, 79, 787–795.[ISI][Medline]
Young, N. (1997) Effects of alcohol and other drugs on children. Journal of Psychoactive Drugs, 29, 23–42.[ISI][Medline]
Received on ; accepted on July 11, 2001
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Petrie, F. Bunn, and G. Byrne Parenting programmes for preventing tobacco, alcohol or drugs misuse in children <18: a systematic review Health Educ. Res., April 1, 2007; 22(2): 177 - 191. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. J. Ross, D. Heim, K. Flatley, J. B. Davies, and M. Sudbery An exploration of street drinking in Drumchapel, Scotland Health Educ. Res., June 1, 2005; 20(3): 314 - 322. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||