A Web-Based, Health Promotion Program for Adolescent Girls and Their Mothers Who Reside in Public Housing

Abstract

This study tested a brief web-based, family-involvement health promotion program aimed at drug use, physical activity, and nutrition for adolescent girls, aged 10 to 12 years, who reside in public housing. Separately, girls (n = 67) and their mothers (n = 67) completed baseline measures online. Following baseline, 36 randomly assigned mother-daughter dyads jointly completed a 3-session, health promotion program online. Subsequently, all girls and mothers separately completed posttest and 5-month follow-up measures. Attrition at posttest and 5-month follow-up measures was 3% and 9%, respectively. At posttest, intervention-arm girls, relative to control-arm girls, reported greater mother-daughter communication and parental monitoring. Intervention-arm mothers reported greater mother-daughter communication and closeness as well as increased vegetable intake and physical activity. At 5-month follow-up, intervention-arm girls and mothers, relative to those in the control arm, reported greater levels of parental monitoring. Intervention-arm girls also reported greater mother-daughter communication and closeness, reduced stress, greater refusal skills, and increased fruit intake. Findings indicate the potential of a brief, web-based program to improve the health of low-income girls and their mothers.

1. Introduction

Adolescent girls who live in public housing are at risk for poor health outcomes owing to their socio-economic status. Low-income youths have poorer nutritional intake and are less physically active than their higher socio-economic peers (Regan, Lee, Booth, & Reese-Smith, 2006). Neighborhood disadvantage may also increase girls’ risks for drug use (Leech, 2012). Because negative health behaviors share common risk factors, health promotion programs must target multiple outcomes (Flay, 2002). And, because for girls in particular, a healthy and close parent-child relationship is associated with positive adolescent development (Chan, Kelly, & Toumbourou, 2013; Kelly et al. 2011; Lombe, Nebbitt, & Mapson, 2009), health promotion programs should involve parents. Programs should also be engaging, simple to use, and easy to disseminate.

Web-based interventions are ideal for meeting the aforementioned demands of health promotion programs. They use a delivery platform familiar to teens and their parents, can be culturally tailored, and require no staff training. Web-based programs also mitigate the logistical challenges commensurate with family-involvement interventions because parents can receive the intervention at home.

Building off our prior work with computer-based interventions (Schinke, Fang, & Cole, 2009; Schwinn & Schinke, 2010; Schwinn, Schinke, & Di Noia, 2010) this pilot study tested a brief, health promotion program for low-income girls and their mothers. The program aimed to enhance the well-being of girls residing in public housing by improving dietary intake, increasing physical activity, and reducing drug use risks.

2. Materials and Methods

2.1. Sample

Participants were from 27 U.S. states and lived in publicly subsidized housing. Across 4 weeks, we ran recruitment ads on Facebook, craigslist, and Google AdWords and in public housing development newspapers. Ads directed mothers to the study’s information website. There, interested mothers requested receipt of the study information packet (an introductory letter, a list of frequently asked questions, enrollment forms, and a franked return envelope) by providing their name and mailing address. We mailed out 86 information packets and received 67 signed youth assent, parent permission, and parent consent forms.

2.2. Procedures

Mother-daughter dyads were randomly assigned to the intervention or control arm. Girls and mothers completed their respective baseline measures online. Subsequently, intervention-arm dyads completed the 3-session health promotion program; control-arms dyads received no intervention materials. All dyads completed posttest and 5-month follow-up measures. Each girl and mother received $20 for baseline, $30 for posttest, and $30 for 5-month follow-up. The Columbia University IRB approved our Human Subjects’ procedures.

2.3. Health Promotion Program

After baseline, intervention-arm mothers and daughters were directed to complete the first session of the health promotion program on a secure website. Working together at the same computer, mothers and daughters completed each session in roughly 25 minutes. Dyads completed one session per week for 3 weeks. E-mail reminders and telephone calls prompted participants when they fell behind schedule.

The integrated, 3-session program focused on developing and maintaining girls’ and mothers’ healthy relationships, bodies, and minds. Session 1 focused on active listening, communication, and the benefits of family meals. In session 2, mothers and daughters discussed their knowledge about drugs. Mothers learned to set and enforce rules. Dyads learned strategies to make healthy and economical decisions at the grocery store and how to make healthy dinners. Session 3 dealt with coping skills. Drawing from a list of stressors that reflect the hardships of living in public housing, dyads identified and shared with each other sources of stress. Mothers and daughters were also exposed to a 5-step problem solving process.

2.4. Fidelity

To assess program completion fidelity, research staff sent mothers and daughters the link to the study website and assigned each dyad member a unique code. Neither member of the dyad was provided with the other member’s code. Separately, with their codes, both members of the mother-daughter dyad concurrently logged onto the website. Only by entering both codes could mother-daughter dyads unlock the health promotion sessions. Session completion was tracked by a real-time backend database linked to the online program. According to tracking data, of the 36 mother-daughter dyads assigned to the health promotion program, 35 completed all three sessions and 1 dyad completed two of the three sessions.

2.5. Measures

2.5.1 Daughter and Mother Measures

Demographic questions asked girls for their age, grade in school, ethnic-racial identification, and average letter grade earned in school. Demographic questions asked mothers for their birthdays, ethnic-racial identification, highest level of education attained, employment status, and whether their daughters qualified for free or reduced price lunch.

Mother-daughter closeness was evaluated with 10 items adapted from the Parental Bonding Instrument (Parker, Tupling, & Brown, 1979). Mothers rated such items as “My daughter likes to talk to me” and “I care about my daughter” with a 6-point Likert scale (“strongly agree” = 1 to “strongly disagree” = 6). Parallel questions were posed to girls. Reliability of the mother and daughter items ranged from α = .73 to .87; the correlation between mothers’ and daughters’ scores across all three time points ranged from r = .65 to .73.

Mother-daughter communication was assessed with six items from the Barnes and Olson Communication Scale (Barnes & Olson, 1985). With a 6-point Likert scale (“strongly agree” = 1 to “strongly disagree” = 6), daughters rated such parameters as “It is easy to discuss my problems with my mother;” and “My mother sometimes doesn’t listen to me.” Parallel questions were asked of mothers but reworded to assess communication quality. Reliability of the mother and daughter items ranged from α = .83 to .86; the correlation between mothers’ and daughters’ scores across all three time points ranged from r = .45 to .60.

Parental monitoring was assessed with five items from the Parental Monitoring Scale (Small & Kerns, 1993). With a 5-point Likert scale (“never” = 1 to “always” = 5), mothers rated such parameters as “I know where my daughter is after school” and “I know my daughter’s friends.” The same questions were reworded and asked of the daughters. Reliability of the mother and daughter items ranged from α = .84 to .95; the correlation between mothers’ and daughters’ scores across all three time points ranged from r = .43 to .56.

Substance use items were adapted from the CDC’s Youth Risk Behavior Survey (YRBS; Centers for Disease Control and Prevention, 2005). Girls reported past-month and past-week use of alcohol, cigarettes, marijuana, heroin, inhalants, methamphetamines, amphetamines, ecstasy, and tranquilizers. Mothers reported past-month and past-week alcohol and cigarette use. Response categories let respondents enter the exact number of times they used a particular substance. Test-retest reliability for YRBS items ranges from .82 to .95 (Centers for Disease Control and Prevention, 2004).

Fruit and vegetable intake was evaluated with 21 items from the Youth and Adolescent Food Frequency questionnaire (Rockett et al., 1997). Mothers and daughters reported how often they consumed certain foods per week (“0 times per week” = 1 to “more than 7 times per week” = 6). The reliability for the fruit items is α = .72 (Speck, Bradley, Harrell, & Belyea, 2001); reliability for the vegetable items is α = .83 (Speck et al., 2001).

Physical activity was evaluated with 12 items from the Kaiser Physical Activity Survey (Ainsworth, Sternfeld, Richardson, & Jackson, 2000). Eight of the items evaluated the types of activities girls and mothers typically engaged in each week. These eight items used a 5-point Likert scale (“0 times per week” = 1 to “more than 7 times per week” = 6). The other four items assessed the number of hours that girls and mothers spent on such activities as “watching TV” or “surfing the Internet” (“less than 1,” “1,” “2,” “3,” or “more than 3 hours per week”). The Kaiser Physical Activity Survey has intra-class correlation coefficients ranging from .79 to .91 (Lee, Im, & Chee, 2009).

2.5.2. Daughter-only Measures

Perceived stress was reported by girls with six items based on the Perceived Stress Scale (Cohen, Kamarck, & Mermelstein, 1983). With a 5-point Likert scale, girls rated the degree to which they felt stressed because of such situations as “feeling fat or unattractive” or “feeling pressure from friends” (“all the time” = 1 to “never” = 5). The reliability of the Perceived Stress Scale is α =.84 (Augustine et al., 2011).

Drug refusal skills were measured with two items that assessed the ease with which girls felt they could refuse an offer to use drugs from a best friend (Fearnow-Kenney, Hansen, & McNeal, Jr., 2002; Hansen & McNeal, Jr., 2001). An illustrative item is “Pretend your best friend offered you marijuana and you did not want it. How hard would it be for you to refuse the offer?” Response options ranged from “very easy =” 1 to “very hard” = 4. Reliability for the scale is α = .84 (Fearnow-Kenney et al., 2002; Hansen & McNeal, Jr., 2001).

2.6. Data Analysis

Analysis of covariance assessed program effects at posttest, controlling for baseline scores. General linear model repeated-measures examined the time by intervention interaction effects at 5-month follow-up. Outcome variables (i.e., mother-daughter closeness, mother-daughter communication, parental monitoring, refusal skills, perceived stress, substance use, fruit intake, vegetable intake, and physical activity) were the within-subject factor and assignment to study arm was the between-subject factor. When sphericity violations occurred, the Greenhouse-Geiser corrected epsilon value was applied to adjust subsequent repeated-measures ANOVAs. Overall attrition was 9% (14% in the intervention arm; 3.2% in the control arm).

3. Results

3.1. Baseline

Sample demographics from baseline data appear in Table 1. Girls and mothers had a mean age of 11.85 years (SD = 0.88) and 36.24 years (SD = 6.16), respectively. Girls were White (43.3%), Black (40%), Latina (13.3%) and Asian (3.3%). Most girls qualified for free (78%) or reduced price (15%) lunch. Intervention- and control-arm participants did not differ in relation to demographic or outcome variables and demographic variables were not associated with outcomes variables.

Table 1.

Demographic Characteristics of Study Participants at Pretest (N = 134)

Demographic variable	All	Control	Intervention	t or χ2	p
Daughters	n = 67	n = 31	n = 36
Age, M (SD)	11.85 (0.88)	11.87 (0.96)	11.83 (0.81)	t(65) = 0.17	.86
Ethnicity, % (n)				χ2(3) = 1.19	.76
White	40.3 (27)	35.5 (11)	44.4 (16)
Black	44.8 (30)	45.2 (14)	44.4 (16)
Latina	11.9 (8)	16.1 (5)	8.3 (3)
Asian	3.0 (2)	3.2 (1)	2.8 (1)
Eligible for school lunch program, % (n)				χ2(2) = .80	.67
Free	79.1 (53)	80.6 (25)	77.8 (28)
Reduced	14.9 (10)	16.1 (5)	13.9 (5)
Does not qualify	6.0 (4)	3.2 (1)	8.3 (3)
Mothers	n = 67	n = 31	n = 36
Age, M (SD)	36.24 (6.16)	36.03 (5.94)	36.43 (6.43)	t(64) = −0.26	.80
Education, % (n)				χ2(4) = 7.12	.13
Some high school of less	7.5 (5)	3.2 (1)	11.1 (4)
Completed high school	31.3 (21)	32.3 (10)	30.6 (11)
Vocational school or 2-year college	32.8 (22)	41.9 (13)	25.0 (9)
4-year college	25.4 (17)	16.1 (5)	33.3 (12)
Graduate school	3.0 (2)	6.5 (2)	0
Employment, % (n)				χ2(3) = 3.16	.37
Full-time	44.8 (30)	41.9 (13)	47.2 (17)
Part-time	16.4 (11)	16.1 (5)	16.7 (6)
Homemaker	11.9 (8)	6.5 (2)	16.7 (6)
Student or unemployed	26.9 (18)	35.5 (11)	19.4 (7)

3.2. Posttest and 5-Month Follow-up

At posttest and relative to the control arm (Table 2), girls and mothers who received the health promotion program reported greater mother-daughter communication (p < .01; p < .05, respectively). Intervention-arm girls also reported more parental monitoring (p < .01). Intervention arm mothers, reported greater closeness to their daughters (p < .05), increased vegetable consumption (p < .05), and increased physical activity (p < .05). Past-month substance use rates did not differ between intervention arm and control arm participants.

Table 2.

Pretest, Posttest, and 5-Month Follow-Up Data for Daughters and Mothers

Outcome variable	Control			Intervention			Pretest to posttesta	Pretest to post-test and 5-month follow-upb
	Pretest (n = 31)	Posttest (n = 31)	5-month follow-up (n = 30)	Pretest (n = 36)	Posttest (n = 35)	5-month follow-up (n = 31)
	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	F (1, 58)	F (2, 58)
Daughter measures
Mother-daughter closeness	4.17 (0.85)	4.00 (1.25)	3.86 (1.33)	3.90 (0.98)	4.03 (0.80)	4.29 (0.82)	0.41	3.18*
Mother-daughter communication	4.17 (1.20)	3.93 (1.19)	4.00 (1.34)	4.10 (1.14)	4.52 (0.63)	4.52 (0.63)	8.33**	4.04*
Parental monitoring	4.76 (0.64)	4.31 (0.93)	4.62 (0.78)	4.48 (0.77)	4.61 (0.56)	4.68 (0.54)	8.89**	6.53*
Perceived stress	4.10 (1.18)	4.14 (0.92)	3.72 (1.41)	3.94 (1.26)	3.94 (0.93)	4.13 (0.72)	0.47	4.32*
Drug refusal skills	3.87 (0.44)	3.83 (0.38)	3.83 (0.38)	3.58 (0.81)	3.48 (1.06)	3.94 (0.25)	1.44	3.49*
30-day substance usec	0.37 (0.91)	0.17 (0.37)	0.10 (0.22)	0.53 (2.19)	0.11 (0.29)	0.12 (0.29)	0.78	0.28
Fruit intake	2.86 (1.66)	2.66 (1.47)	2.62 (1.24)	2.39 (1.17)	2.35 (1.14)	2.90 (1.11)	0.02	3.61*
Vegetable intake	2.31 (0.91)	2.51 (1.08)	2.20 (0.94)	2.28 (0.90)	2.38 (0.81)	2.41 (0.99)	0.33	2.02
Physical activity	2.29 (0.62)	2.41 (0.70)	2.25 (0.71)	2.17 (0.65)	2.19 (0.70)	2.12 (0.71)	0.58	0.21
Mother measures
Mother-daughter closeness	4.80 (0.93)	4.80 (0.55)	4.83 (0.75)	5.00 (0.00)	5.00 (0.00)	4.97 (0.18)	4.34*	0.90
Mother-daughter communication	3.67 (0.80)	3.68 (1.17)	3.77 (1.140	3.58 (1.06)	4.06 (0.87)	3.94 (0.81)	4.54*	1.63
Parental monitoring	4.70 (0.65)	4.60 (0.86)	4.47 (1.04)	4.42 (1.03)	4.52 (0.85)	4.77 (0.50)	0.27	3.58*
30-day alcohol use	3.66 (5.82)	2.59 (4.08)	1.52 (2.63)	2.74 (5.59)	2.26 (5.67)	1.65 (2.63)	0.05	0.36
30-day cigarette use	4.55 (14.46)	5.00 (14.44)	3.69 (13.38)	1.32 (5.53)	0.32 (0.54)	0.26 (0.45)	1.39	1.18

Note.

^aAnalysis of covariance (ANCOVA) results, controlling for pretest outcome variable scores.

^bTime by intervention interaction effect based on repeated-measures general linear modeling results.

^cIndex variable for scores on 30-day alcohol, cigarettes, and marijuana.

^*p < .05.

^**p < .01.

At 5-month follow-up (Table 2), time by intervention interaction results showed that over time, intervention-arm girls demonstrated greater mother-daughter communication (p < .05), closeness (p < .05), increased fruit consumption (p < .05), reduced psychosocial stress (p < .05), and greater drug use refusal skills (p < .05), relative to control-arm girls. Intervention-arm girls and mothers reported increased parental monitoring (p < .01; p < .05, respectively). Again, there were no differences in past-month substance use between the intervention- and control-arm participants.

4. Discussion

Building off our prior work with computer- and Internet-based studies (Schinke, Fang, & Cole, 2009; Schwinn & Schinke, 2010; Schwinn, Schinke, & Di Noia, 2010), these findings are the first of their kind from a health promotion program expressly aimed at adolescent girls and their mothers living in public housing. Findings suggest that a brief, web-based health promotion program for such girls and their mothers can affect positive and relatively sustained changes in health behavior and salient risk and protective factors. Girls and mothers improved their scores on measures of communication, closeness, and parental monitoring. Girls increased their consumption of fresh fruit, and mothers increased their physical activity and consumption of vegetables. That girls reported better scores on measures of stress and drug refusal responses 5 months after they received the health promotion program lends credence to the program’s continuing effects. The modest correlation between mothers’ and daughters’ scores on closeness, communication, and parental monitoring suggests that though mother-daughter scores are somewhat predictive of each other (R² = 18% to 53%), mothers’ and daughters’ perceptions of their relationship differ.

Limitations include a brief intervention, a short follow-up, and a small sample. An intervention that imparts a greater constellation of skills and enhanced levels of interactivity is warranted. Additional follow-up periods should assess long-term impact. And, the sample, though coming from over one-half of the states in America, was limited in size and included mothers with impressive educational attainment—i.e. 28.4% had a 4-year college degree or higher. Follow-on work should also monitor participant receipt of other health promotion programs – something this study did not do. Beyond these improvements, tailored health promotion programs should exploit the advantages of computer programming to build programs adaptable to multiple family relations (e.g., mothers and sons, fathers and sons). Still, findings point toward the benefits of focusing on girls’ and mothers’ overall health by integrating skills related to improving a healthy body, mind, and parent-child relationship.

Highlights.

• Developed and tested a brief, family- and web-based health promotion program.

• Designed for adolescent girls and their mothers who reside in public housing.

• Focused on dietary intake, physical activity, and substance use risks.

• Findings indicate the potential to improve multiple health behaviors.