Randomized Trial of a Parent Intervention: Parents Can Make a Difference in Long-term Adolescent Risk Behaviors, Perceptions, and Knowledge

Background  Although numerous interventions have been demonstrated to reduce targeted adolescent risk behaviors for brief periods, sustained behavior changes covering multiple risk behaviors have been elusive.

Objective  To determine whether a parental monitoring intervention (Informed Parents and Children Together [ImPACT]) with and without boosters can further reduce adolescent truancy, substance abuse, and sexual risk behaviors and can alter related perceptions 24 months after intervention among youth who have all received an adolescent risk-reduction intervention, Focus on Kids (FOK).

Design  Randomized, controlled, 3-celled longitudinal trial.

Setting  Thirty-five low-income, urban community sites.

Participants  Eight hundred seventeen African American youth aged 13 to 16 at baseline.

Intervention  All youth participated in FOK, an 8-session, theory-based, small group, face-to-face risk-reduction intervention, 496 youth and parents received the 1-session ImPACT intervention (a videotape and discussion), 238 of the ImPACT youth also received four 90-minute FOK boosters delivered in small groups.

Main Outcome Measures  Responses at baseline and 24 months after intervention to a questionnaire assessing risk and protective behaviors and perceptions. Analyses used General Linear Modeling, intraclass correlation coefficient, analysis of covariance, and multiple comparisons with least significant difference test adjustment.

Results  After adjusting for the intraclass correlation coefficient, 6 of 16 risk behaviors were significantly reduced (P≤.05) among youth receiving ImPACT compared with youth who only received FOK (respectively, mean number of days suspended, 0.65 vs 1.17; carry a bat as a weapon, 4.1% vs 9.6%; smoked cigarettes, 12.5% vs 22.7%; used marijuana, 18.3% vs 26.8%; used other illicit drugs, 1.4% vs 5.6%; and, asked sexual partner if condom always used, 77.9% vs 64.9%). Four of the 7 theory-based subscales reflected significant protective changes among youth who received ImPACT. ImPACT did not produce any significant adverse effects on behaviors or perceptions.

Conclusion  A parent monitoring intervention can significantly broaden and sustain protection beyond that conferred through an adolescent risk-reduction intervention.

The past decade has enjoyed the emergence of several adolescent risk-prevention programs of demonstrated efficacy toward the reduction of sexual risk behaviors,¹ substance abuse,² and tobacco use prevention.¹ Consensus, with some evidence base, provides a template for the shared characteristics of these successful programs: theory-based, practice in skills, attentive to personal values and social norms, narrowly focused on specific risk behaviors, and using multiple delivery formats.^3-5 The process of developing and evaluating these efforts has led to a richer understanding of adolescent behavior and its determinants.⁶ The importance of peer influence has been consistently demonstrated in multiple studies.^7-10 The continued importance of parental influences has emerged as an equally strong determinant of not only just childhood but also adolescent behavior.^10-14

As substantial as these efforts have been, their transient influence on risk behavior has been disappointing for those interventionists who have attempted to follow up youths through the adolescent years. Few interventions have been assessed beyond 12 months; among those which have, most have found negligible effects.¹⁵

To test the hypothesis that the reinforcing effects of peer networks might alternatively strengthen or weaken an intervention effect, almost a decade ago we elected to provide a risk-reduction intervention, “Focus on Kids” (FOK), to naturally occurring peer (rather than investigator-formed) groups.¹⁶ While the short-term effects of this intervention approach were significant and, after boosters, there was a resurgence of intervention effect at 18 months, intervention effect waned over time.¹⁷ We observed that the friendship groups were not stable over time and speculated that this breakdown of reference group may have contributed to the lessening effect of the intervention.

Different from friends, parents are, more or less, permanent in an adolescent’s life. As our data and that of other investigators suggest,^13,18,19 parents retain their roles as important determinants of risk and protective behaviors among adolescents, competing admirably with the influence of peers.^10,14 The role of monitoring (defined as parental communication and supervision of their children) seems to be especially important in reducing risk and increasing protective actions.^13,20 We, therefore, hypothesized that a parental-monitoring intervention might sustain—and perhaps broaden—intervention effects. In a preliminary study, we demonstrated that a parental-monitoring intervention, “Informed Parents and Children Together” (ImPACT), did increase parent-youth communication and perceptions of parental monitoring; although as the sole intervention, it did not reduce adolescent self-reports of risk behavior.²¹ Intriguingly, intervention effect on parent-child communication increased over time, supporting the aforementioned hypothesis that for interventions addressing skill sets, effects might be augmented rather than diminished over time.

Given the ethical imperative of the state of uncertainty or “equipoise” to justify a control or placebo group in a clinical trial²² and the demonstrated efficacy of FOK and similar adolescent risk-prevention interventions in our target population,¹⁶ we could not justify a trial using a control group against adolescent risk behavior. However, we could offer FOK to all youth and augment it with a parental-monitoring intervention, ImPACT, to determine if this intervention component sustained and/or broadened the intervention effect. We hypothesized that in the short-term (6 months) and moderate-term (12 months), intervention effects would be stronger in youth assigned to FOK combined with ImPACT and include multiple risk behaviors (as opposed to only sexual risk behaviors, the primary target of FOK). Indeed, we have reported that among youth who received FOK, those youths whose parents had been randomized to receive ImPACT demonstrated reduced rates of sexual risk behaviors and substance use at 6 months and reduced rates of substance use and overall risk intent at 12 months.²³ These results left unanswered whether such effects would be robust across time and across normal adolescent developmental changes. We report herein on the long-term (24-month) results with regard to behaviors and putatively influential perceptions.

This randomized, longitudinal trial was conducted among 817 youth located in 35 low-income urban sites in Baltimore, Md. Youths were recruited over 3 waves in 1999 and 2000; wave 1 included youth from 8 sites, wave 2 from 10 sites, and wave 3 from 17 sites. Randomization occurred at the level of the site. As given in Table 1, 39% of youth were randomized to receive FOK only (eg, youth received FOK and parents received the attention-control “Goal for IT”), 32% received FOK + ImPACT (without FOK boosters), and 29% received FOK + ImPACT + boosters. The median age of the group was 14 years and 58% were female; intervention groups were similar for sex and age at baseline. Follow-up assessments were conducted at 6, 12, 18, and 24 months after intervention. The research was approved by the institutional review board at the University of Maryland, Baltimore. Written, informed consent/assent was obtained from parents and youths.

Recruitment sites were selected to access the youths living in low-income communities including all public housing development. Housing development tenant association members and local recreation center staff were invited to work as community recruiters. These recruiters assisted in establishing strategies for identifying and recruiting eligible youth and parents within their particular community. Randomization occurred at the level of the 35 sites and was conducted by a random numbers table after all youths at all sites within the recruitment wave had been identified. Thirteen sites were randomized to receive FOK only, 11 to FOK+ImPACT, and 11 to FOK+ImPACT+ boosters.

The local facilitators identified eligible youths, described the program to the youths and their parents, and established an appointment time for enrollment and baseline data collection from those who were interested. Appointments occurred in the youth’s home or at a designated community site. During the initial appointment, the youths and their parent completed assent/consent forms and a baseline questionnaire in separate rooms. Immediately following completion of baseline surveys, parents and youths participated in either the ImPACT program or the parent attention-control according to the predetermined randomization status of the community. The interventionist closed the appointment by giving the youth and parent information about the FOK sessions scheduled for the youth. Although youths were eligible to enroll even if their parent or guardian was unwilling to participate in the study, all invited parents did participate in their assigned interventions.

The interventions in this study were based on a social cognitive model, Protection Motivation Theory (PMT).²⁴ According to PMT (Figure 1), environmental and personal factors combine to create a potential health threat. The perceived threat initiates 2 cognitive pathways, a threat-appraisal pathway and a coping-appraisal pathway. The threat-appraisal pathway evaluates the factors associated with the threat including perceived intrinsic and extrinsic rewards accompanying the behavior minus the perceived severity of the threat and one’s vulnerability to the threat. The coping-appraisal pathway evaluates one’s ability to avert the threatened danger including both self-efficacy and efficacy of the response balanced against the cost of the response. These 2 appraisal pathways combine to produce protection motivation, which, if high enough, may result in protective action.

Because the 3 intervention components (FOK, ImPACT, and FOK boosters) have been described in detail previously,^16,17,21 only a brief summary is provided herein. Focus on Kids is an 8-session (each approximately 1.5 hours) risk-reduction intervention that emphasizes decision making, goal setting, communicating, negotiating, and consensual relationships and information regarding abstinence and safe sex, drugs, alcohol, and drug selling. Intervention format includes games, discussions, homework assignments, and videotapes and is delivered to groups of 5 to 12 youth by a pair of interventionists. ImPACT includes a 20-minute videotape (designed for the targeted communities) emphasizing several concepts of parental monitoring and communication. ImPACT is delivered by a pair of interventionists through a portable videotape player and is followed by an interactive role-play describing a confrontational scenario. After the role-play between the parent and the youth is completed, the interventionist critiques it according to the main talking points of the videotape and conducts a condom demonstration.

For ImPACT communities that were randomized to the FOK booster sessions’ arm of the study, youths were invited to a 90-minute booster session that occurred at the same community site where the original FOK program was completed. If an invited youth did not attend the booster session, an attempt was made to complete the booster session during a home visit with the youth. The booster sessions, delivered at 7, 10, 13, and 16 months after intervention, reviewed material regarding decision making and communicating regarding sexual and other risk-protective behaviors, and include both repeat activities from the original sessions as well as new activities emphasizing the same principles and content. The attention control (“Goal for IT”) for ImPACT consists of a 20-minute videotape describing the process for establishing and implementing career goals, followed by a brief discussion scripted by a written text. Control boosters were not offered.

Youth risk behaviors were assessed by youth self-report, using the Youth Health Risk Behavior Inventory. Originally developed for urban African American youth in Baltimore,²⁵ the first section of the Youth Health Risk Behavior Inventory assesses demographic characteristics of the youth. The second section assesses the youth's involvement in risk behaviors including delinquent behaviors, drug-related behaviors, and sexual risk behaviors during the previous 6 months. Dichotomous responses (0, no; 1, yes) were used for these items. In the next section, youths are queried along a 5-point Likert scale about their perceptions of risk and protective behaviors according to the 7 constructs of PMT. The Youth Health Risk Behavior Inventory questions were designed based on qualitative research and theoretic considerations and were subsequently subjected to reliability testing and assessment of internal consistency (Cronbach α), face validity (a type of commonsense appraisal that is beyond any mathematical formulas), and criterion-related validity.²⁵ In this study, because we are interested in the overall effect of the interventions on the broad area of risk perception, the specific risk scales were collapsed into single PMT construct subscales. Potential items were backed out of the construct subscale in an effort to reach or exceed an α value of 0.70. α Values exceeded 0.70 for 6 of the subscales (exceeding 0.80 for 3 subscales) with the seventh, response efficacy, having an α of 0.69. Consistent with PMT, higher mean values on the self-efficacy, response efficacy, severity, and vulnerability subscales and lower values for intrinsic rewards, extrinsic rewards, and response cost subscales are protective.

The questionnaires were administered orally and visually by computer. This method is described in greater detail elsewhere.²⁶ Questionnaires required approximately 45 minutes to administer.

We assessed baseline equivalence of demographic characteristics using χ² and Kruskal-Wallis tests. Baseline differences in behaviors and perceptions among the different intervention groups (ie, FOK only, FOK + ImPACT, and FOK + ImPACT + boosters) were assessed using the General Linear Model. Any baseline variables that differed significantly among the groups were controlled for in subsequent analysis of covariance analyses of intervention effect on perceptions, knowledge, and behaviors.

Three types of intervention effects at 24 months were assessed among the intervention groups: the overall intervention effect (ie, difference among FOK only vs FOK + ImPACT, and FOK only vs FOK + ImPACT + boosters); the overall additive effect of parental intervention to the FOK (difference between FOK only and FOK + ImPACT with/without boosters); and, the additive effect of FOK boosters to the parental intervention (difference between FOK + ImPACT groups by the presence or absence of boosters). We used the least significant differences (LSDs) multiple comparisons procedure to control for type I error.²⁷

To adjust for the strategy of cluster (rather than individual) randomization, the intraclass correlation coefficient (ICC) was determined for each behavior and construct subscale score (and are contained within Table 2, Table 3, and Table 4). A corrected independent sample t test, which was adjusted for the ICC,²⁸ was performed to adjust the test statistics for the group difference.

As given in Table 1, at 24 months after intervention, follow-up rates were comparable among intervention groups, proportionate to the original intervention assignments as outlined earlier in the “General” subsection of the “Methods” section. From the 817 youth at baseline, 494 (60%) were present at 24 months; distribution across intervention groups among those present at 24 months was proportionate to the original assignment among the full (baseline) study population. Baseline data on youths absent at 24 months compared with those present reveal that the former were older, more likely to be male, and had higher rates of risk behaviors. However, the baseline demographic and risk-protective behaviors of youths absent at 24 months were comparable across intervention groups. That is, despite the dropout rate, the baseline risk profile remained equivalent across intervention groups.

In Table 2, we display the 24-month results according to intervention status for behaviors. Data are presented for the 16 main risk-protective behaviors assessed in the Youth Health Risk Behavior Inventory, including 4 delinquent, 5 substance use, and 7 sexual risk behaviors. In the first 5 columns of the table, we display the ICC for each behavior, followed by the mean value for youth who received each of the intervention conditions. In the next 4 columns we display the significance level of the pairwise differences among the various intervention conditions, after adjusting for ICC (when the ICC>0.00).

After adjusting for the ICC, 6 of the behaviors differed significantly on the basis of intervention assignment to FOK only vs FOK + ImPACT(with or without boosters) and in all cases risk behaviors were lower among those youth whose parents had been assigned to ImPACT. Specifically, 2 of the general risk behaviors (days suspended and carrying a bat as a weapon), 3 of the substance abuse behaviors (use of cigarettes, marijuana, and other illicit drugs), and 1 of the sexual risk behaviors (partner notification) differed according to randomization to FOK only vs FOK + ImPACT youth. The difference in 2 additional sexual behaviors were marginally significant (P ≤ .10) and again favored the addition of ImPACT.

The difference of 2 behaviors between FOK only and FOK + ImPACT youth were significant or marginally significant; in all cases, risk behaviors were lower among the ImPACT-enhanced group. Seven behaviors differed in a protective fashion (significantly or marginally significant) among those who had also received FOK + ImPACT + boosters compared with FOK only. Finally, 3 behaviors differed (siginificantly and marginally significant) between the 2 ImPACT groups. For 2 of these behaviors, the addition of boosters offered further significant protection. For 1 behavior, the addition of the booster was associated with greater risk.

Table 3 and Table 4 display perceptions of risk and protective behaviors at the 24-month follow-up according to intervention status and organized by the constructs in the 2 PMT pathways, coping-appraisal and threat-appraisal (see the “Guiding Model of Behavioral Change” subsection of the “Methods” section). For each construct subscale, the construct name, ICC, Cronbach α value for the subscale, and mean values by the 2 main intervention groups (FOK only vs FOK + ImPACT with or without FOK boosters) and significance thereof (after adjusting for ICC where ICC>0.00) are indicated in boldfaced type, below which are the items composing that subscale.

Overall, as shown in Table 3, Cronbach α values exceeded 0.80 for 2 of the subscales and was 0.69 for the third. Mean self-efficacy values were significantly higher among youth who received FOK + ImPACT compared with youths who received FOK only. Of the 20 items included in the self-efficacy subscale, 9 demonstrated a significant difference based on intervention assignment; all differences reflected increased perceptions of self-efficacy among youths receiving FOK + ImPACT with or without FOK boosters.

The mean values of the response efficacy subscale also demonstrated a significant protective effect among youths receiving FOK + ImPACT compared with youth receiving FOK only. Of the 13 items composing the subscale, 1 item differed significantly, reflecting enhanced protection through the addition of ImPACT to FOK.

The final coping-appraisal construct, response cost, demonstrated marginally significant enhancement of protection (eg, lowered perceptions of response cost for enacting the protective behavior) among youths who received FOK + ImPACT compared with youths who received FOK only. Not shown in this table, youths receiving FOK + ImPACT without the FOK boosters demonstrated significantly lower perceptions of response cost compared with FOK only youths (54.08 vs 57.48, P = .02). Among the 21 items in the response cost subscale, 8 perceptions differed significantly, 6 of which demonstrated protection conferred from ImPACT.

The α values exceeded 0.70 for all 4 subscales assessing this pathway. Only 1 subscale, intrinsic rewards, demonstrated a significant difference in mean values between intervention groups; values were significantly lower (eg, showed a protective effect) among youths receiving FOK + ImPACT (with and without FOK boosters) compared with youths receiving FOK only. Of the 10 items included in the subscale, 3 showed a statistically significant difference, all of which reflected enhanced protection among youths receiving ImPACT. The overall scores for the other 3 subscales (extrinsic rewards, vulnerability, and severity) did not differ significantly between intervention groups.

These results demonstrate the ability of a supplemental intervention among parents to enhance and sustain the intervention effect enjoyed from an adolescent risk-reduction intervention. The results are consistent with the substantial data that demonstrate the importance of parents throughout the adolescent years.^13,18 In some instances, the addition of the FOK boosters also conferred added protection. Both the parent intervention and the FOK booster intervention are consistent with developmental theory that suggests that youths will need to reassess situations as they age and assimilate new experiences.^29,30 This finding offers some respite from the polarizing arguments of the last decade over the relative roles of parents vs extrafamilial interventions in educating children about sexual and other risk-reduction efforts.^31-34 Parent and community sources of information can be reinforcing and even augmenting.

The strong intervention effect on the coping-apprai sal pathway, with a less consistent effect on the threat-appraisal pathway, is also consistent with the literature. Perceptions of self-efficacy and their relation to enactment of a wide array of protective maneuvers have been a relatively robust finding in the literature^35-37 as have been perceptions of response cost as motivators of both risky and protective behaviors.^38,39 Constructs from the threat-appraisal pathway, in particular vulnerability, have a long history of conflicting, at times apparently contradictory, associations. Thus, for example, both perceptions of vulnerability and invulnerability have been associated with increased participation in risk behavior. Likewise perceptions of severity are problematic, confounded by proximate and distal outcomes that often have little meaning to an adolescent.⁴⁰

First, all youths received a risk-reduction intervention previously demonstrated to be effective (FOK) because we felt that not to do so would have been offering less than standard of care.²² Thus, while we have demonstrated that FOK + ImPACT is superior to FOK only, we are presuming that FOK was better than a control based on past rather than concurrent data. Second, these data are based on self-report, without biologic or other confirmation. Third, while these data suggest that in some situations the boosters may have enhanced protection, but in others reduced protection, and because we did not have a cell in which only the boosters were added to FOK, no definitive statement can be made regarding their role. Fourth, in enrolling this community-based convenience sample, data were not maintained regarding potentially eligible youths who refused or were not approached. Fifth, there was a substantial attrition of youths at 24 months, whose baseline characteristics indicated greater risk than those youths remaining for follow-up. However, the baseline demographic and risk characteristics did not differ across intervention groups, thus reducing the likelihood that attrition bias influenced the outcome of this study.

ImPACT resulted in multiple significant differences demonstrating protection in long-term outcomes, including behaviors and perceptions that are posited to influence behavior.⁴¹ These findings demonstrate both that a broad range of risk behaviors can be reduced over time and that the underlying factors presumed to be contributing to these changes were altered in a fashion consistent with the observed behavioral changes. These data suggest that properly designed risk-reduction interventions can exert sustained effects over a broad array of behaviors. However, the evidence from this study suggests that to have this effect, interventions will require multiple delivery approaches. In this study, we used face-to-face strategies addressing youths and their parents. Other approaches (including the mass media which might change social norms^9,42) might also be effective, but further research will be needed to assess such combinations.

These data underscore the importance of looking to parents in our efforts to minimize risk exposure among adolescents. Parents really do make a difference in the lives of their adolescents.

Article

The past decade has enjoyed the emergence of several adolescent risk prevention programs of demonstrated efficacy toward the reduction of sexual risk behaviors, substance abuse use, and tobacco use prevention. As substantial as these efforts have been, their relatively transient influence on a narrow array of risk behavior has been disappointing for those interventionists who have attempted to follow up youths through the adolescent years. Intervention efforts that sustain and broaden risk prevention effects among adolescents are needed.

This study demonstrates the ability of a supplemental intervention among parents to substantially enhance and sustain the intervention effect enjoyed from an adolescent risk-reduction intervention. Intervention effect on behaviors, perceptions and knowledge was demonstrated 2 years after intervention.

Correspondence: Bonita Stanton, MD, Department of Pediatrics, Children’s Hospital of Michigan, Suite 1k40, Wayne State University, 3901 Beaubien St, Detroit, MI 48201 (bstanton@dmc.org).

Accepted for Publication: April 22, 2004.

Funding/Support: This study was supported by grant R01MH54983 from the National Institute of Mental Health, Bethesda, Md.

Acknowledgments: We thank the many community members and leaders, interventionists, data collectors, other faculty and staff who worked with us on this project, especially Yvonne Summers. We thank the University of Maryland and West Virginia University for their support of our efforts. Finally, we thank Philtrice Ervin for help in preparing the manuscript.