Address reprint requests to Cheryl L. Perry, PhD, Division of Epidemi- ology, School of ..... Farris RP, Frank GC, Webber LS, Berenson GS: A group method for.
Parent Involvement with Children's Health Promotion: The Minnesota Home Team CHERYL L. PERRY, PHD, RUSSELL V. LUEPKER, PHD, DAVID M. MuRRAY, PHD, CANDACE KURTH, REBECCA MULLIS, PHD, SUSAN CROCKETT, PHD, AND DAVID R. JACOBS, JR., PHD
MPH,
Abstract: This study compares the efficacy of a school-based program to an equivalent home-based program with 2,250 third grade students in 31 urban schools in Minnesota in order to detect changes in dietary fat and sodium consumption. The school-based program, Hearty Heart and Friends, involved 15 sessions over five weeks in the third grade classrooms. The home-based program, the Home Team, involved a five-week correspondence course with the third graders, where parental involvement was necessary in order to complete the activities. Outcome measures included anthropometric, psychosocial and behavioral assessments at school, and dietary recall, food shelf inventories, and urinary sodium data collected in
the students' homes. Participation rates for all aspects of the study were notably high. Eighty-six per cent of the parents participated in the Home Team and 71 per cent (nearly 1,000 families) completed the five-week course. Students in the school-based program had gained more knowledge at posttest than students in the home-based program or controls. Students in the home-based program, however, reported more behavior change, had reduced the total fat, saturated fat, and monounsaturated fat in their diets, and had more of the encouraged foods on their food shelves. The data converge to suggest the feasibility and importance of parental involvement for health behavior changes with children of this age. (Am J Public Health 1988; 78:
Introduction A major dilemma in children's health promotion is how to involve parents in those efforts. Parents serve important health-related roles for their children-as models of appropriate behavior, as gatekeepers to both opportunities and barriers, and as the major sources of reinforcement in most children's lives. Although the influence of parents on children's health behavior is well-documented, the question of how to modify that influence to be more health-enhancing has not been adequately studied. Efforts to involve parents in children's health promotion have not been encouraging, primarily due to the difficulties in recruiting and maintaining substantial parental participation. This study addresses the issue of parent involvement in a children's health promotion program designed to modify eating patterns among third graders (ages 7-9) in 31 urban elementary schools. Children's eating habits in the United States reflect those of adults: a diet that is high in saturated fat and sodium.' These eating patterns may contribute to the increased levels of cardiovascular risk factors among American children,2'3 thus underscoring the need for programs focusing on change in children's eating patterns for cardiovascular health
in nutrition education suggested by this research is to emphasize healthy eating patterns and to teach skills that facilitate the learning and practice of these patterns. Parent involvement appears crucial for the implementation and maintenance of new health behaviors in younger children. 12-4 Parents are important targets for health education efforts because they act as role models and teachers for their children. Health education interventions which change family attitudes and habits are likely to promote longerlasting health behavior changes.15 Still, research on changing health-related behaviors within a family context is limited. "22 Perry, Crockett, and Pirie23 reviewed this literature, reaching three conclusions: * Children are able to influence the attitudes and behaviors of their parents. * While health education classes for parents are a useful intervention for high risk populations, significant nonattendance rates and difficulty of recruitment present barriers to reaching most parents. * Parents prefer more flexible educational methods, such as printed materials or activities that can be completed at home with their children. Thus a nontraditional, non-class format appears most efficacious for a family intervention program. The two educational programs developed for this study test the relative merits of a school-based eating patterns program for third grade children versus a home-based approach for the students and their parents. The programs are based on previous work in cardiovascular health promotion in schools,4"'1 and extend these concepts to the home in a non-traditional, correspondence, activity format.
promotion.4
Reviews of the nutrition education literature suggest that school-based interventions can increase students' knowl-
edge.5'6 However, few programs have been successful in altering children's eating behavior. This may be due to a
limited understanding of children's eating patterns, weak intervention strategies, poor research designs, or outcome measures that are unable to detect changes in behavior.7 In recent years a small number of programs have addressed these issues and achieved limited success in changing behavior in elementary school students.>l" The current direction Address reprint requests to Cheryl L. Perry, PhD, Division of Epidemiology, School of Public Health, University of Minnesota, 611 Beacon Street, SE, Stadium Gate 27, Minneapolis, MN 55455. All co-authors are with the same University and Division, except Dr. Crockett who is with the College of Home Economics, North Dakota State University. This paper, submitted to the Journal December 10, 1987, was revised and accepted for publication March 24, 1988. Editor's Note: See also related editorial p 1149 this issue © 1988 American Journal of Public Health 0090-0036/88$1.50
1156
1156-1160.)
Methods The study employed a pretest-posttest factorial design involving 32 schools in four urban school districts in Minnesota and North Dakota. One school in one school district withdrew from the study due to commitments to a previously developed nutrition education program. The remaining 31 schools were blocked by state and randomly assigned to one of four conditions: the school-based Hearty Heart program (HH), the home-based Home Team program (HT), both programs in sequence (HH, HT), or a no-treatment control AJPH September 1988, Vol. 78, No. 9
PARENT INVOLVEMENT IN CHILDREN'S HEALTH PROMOTION
group (C). The assigned education program was given to all third grade students who were enrolled in the participating schools. All children were pretested in the Spring of 1985 as second graders. The intervention took place in the 1985-86 school year with the posttest in the spring of 1986. The study population was predominantly caucasian and middle class. Program Development for Hearty Heart and the Home Team
The Hearty Heart program is a five-week 15-session school curriculum taught by third grade classroom teachers. The theory used to drive the study was derived originally from social learning theory, but has been modified for health promotion program development based on our previous work.24'25 The program targets changes in specific environmental, personality, and behavioral factors that are likely to influence children's health behavior. The theoretical model, curriculum design, teacher training program, and pilot outcome data for the Hearty Heart program are described in detail elsewhere."1 The program emphasizes food differentiation (between "everyday" and "sometimes" foods), modeling of healthful eating habits by slide-tape cartoon characters (Hearty Heart, Dynamite Diet, Salt Sleuth, Flash Fitness), food selection and preparation skills, and goalsetting with direct reinforcement. Parental involvement is not sought in this program. The Home Team program is a five-week correspondence course involving third graders and their parents. The concept of a correspondence format came from three sources: * The low participation and high drop-out rates found in traditional parent education programs suggest that an alternative is needed.23 * A fifth grade correspondence program, the Planet Strongheart News, had achieved high participation rates with children although it was entirely homebased.26 * A phone survey to a random sample of parents of fourth graders in one of the target areas was undertaken with 95 per cent participation. This survey asked opinions on nutrition and parent education and involved 208 parents (119 fathers and 89 mothers). The survey ranked preferred parent education methods, in this order (from most to least interest): behavioral tip sheets (75 per cent reported interest), worksheets and homework from school that required parent involvement (67 per cent), informational brochures (42 per cent), phone calls on nutrition (39 per cent), and parent education nights (15 per cent). Thus it seemed necessary to develop a home-based program that could utilize behavioral tip sheets and worksheets at home, be accomplished within busy family schedules, and would provide significant external incentives for continued parental involvement. The Home Team correspondence program is equivalent in content and exposure time to the Hearty Heart program and follows a similar theoretical model. The program consists of five packets that are mailed to each third grader's home on a weekly basis. The packets are designed as a family game using a baseball motif. Each packet contains a Player's Guide describing two to three hours of activities concerning eating pattern changes. The activities begin with the child reading a Hearty Heart Adventure Story to the parent. The story provides role models for healthy eating and basic nutritional information. The parent and child then play games to practice the suggested skills. The games involve mix 'n match of "everyday" and "sometimes" food cards, stickers to label AJPH September 1988,
Vol. 78, No. 9
household foods as heart healthy, and sodium searches, among others. Goals are set to change particular eating
habits, generally by introducing fruits, vegetables, complex carbohydrates, low-fat dairy products, and lean meats into their diet. Recipes such as "Fruit, Fruit, Fruit for the Home Team" and "Championship Veggies" are included. The Team Tips, a refrigerator tip sheet, provides more detailed
nutrition information. The child and parent receive "participation points" for completing the activities together. The points are recorded on a scorecard that is returned to the classroom. University personnel, acting as Home Team coaches, visit the classroom weekly for 10-15 minutes. The coaches collect the scorecards, record points on a Home Team Scoreboard, answer questions, and encourage participation. During the program, Home Team hats, Salt Sleuth magnifying glasses, and a variety of stickers are used as incentives to participate. Those who complete the program, as evidenced by the participation points, are eligible for a drawing for a grand prize-a trip for four to Disneyworld, funded by a local foundation. Survey Procedures Outcome evaluation measures were taken in the school and in the home by a trained survey team. At each survey time, students enrolled in the appropriate grade in all 31 schools took part in the school surveys (N 2,250). Thus, a longitudinal cohort was constructed for the anthropometric, behavioral, and psychosocial assessments. Height, weight, and skinfold thickness measures, taken in the school's gymnasium or cafeteria, were used to assess changes in overweightness that might result from a dietary intervention. The psychosocial questionnaire was administered in the classroom. The questionnaire measured skills, label reading, knowledge, and self-reported food selections. Each question and possible responses were read aloud by an interviewer while students individually marked answers on the questionnaire. Most responses were illustrated with a picture as well as the written word. Because the students were only second graders at the time of the pretest, the psychosocial questionnaire was limited to simple constructs. Questionnaire items were used to create indices to measure these constructs. Higher scores are associated with healthy behavior, greater skills, and accurate knowledge. The following variables were defined: knowledge of high sodium foods (Salt 1, Salt 2), knowledge of high fat foods (Fat 1, Fat 2), knowledge of relatively "heart healthy" foods (Know 1, Know 2), label reading skills to identify salt and sodium (Labels), and self-reported food selections (Behavior). The score range for each index is: Salt 1 (0-5), Fat 1 (0-5), Know 1 (0-6), Salt 2 (0-7), Fat 2 (0-9), Know 2 (0-6), Labels (0-4), Behavior (0-20), with a higher score indicating more knowledge, skills, or better behavior. Behavior was measured at posttest only. The Cronbach alpha coefficients range from .25 for knowledge to .63 for the behavior index, with an average of .37. Test-retest correlations over a one-week interval were .82 for the knowledge and .89 for the behavior indices. These psychometric properties suggest that the students' responses are reliable, but have not yet formed consistent scales, except for behavior. At each observation time, two subsets of students were survey. Thirty randomly selected to participate in a home to provide an students from each school were selected from each students additional 15 overnight urine sample; school were selected to provide a urine sample plus a 24-hour dietary recall. Thus, two independent cross-sectional sam=
1157
PERRY, ET AL. TABLE 1-Differences of the Treatment from the Control Group (ANCOVA with pretest as covariate)
Control Group N = 422
Posttest Means for Knowledge, Skills, and Behavior and 95% Confidence Intervals Home Team N = 512
Hearty Heart N = 387
Hearty Heart & Home Team N = 506
Index*
Number of Items
Mean
(95% Cls)
Diff
(95% Cis)
Diff
(95% Cis)
Diff
(95% Cls)
Salt 1 Fat 1 Know 1 Salt 2 Fat 2 Know 2 Labels Behavior
5 6 6 7 9 6 4 20
63.5 76.2 76.5 81.6 67.0 75.0 27.2 48.6
(61.7, 65.4) (74.6, 77.8) (75.0, 78.0) (80.3, 82.9) (65.6, 68.5) (73.6, 76.5) (25.7, 28.7) (47.1, 50.1)
13.1 10.9 6.1 10.1 10.7 12.8 14.7 4.0
(10.5,15.8) (8.6, 13.2) (3.9, 8.3) (8.2,12.0) (8.6, 12.8) (10.6,14.9) (12.6,16.8) (1.8, 6.2)
9.4 5.3 4.9 5.3 7.5 8.3 8.3 6.4
(6.9, 11.9) (3.1, 7.5) (2.9, 6.9) (3.5, 7.1) (5.6, 9.5)
17.5 11.3 8.0 8.5 9.7 13.7 13.2 8.7
(15.0, 20.0) (9.1, 13.5)
(6.3,10.2) (6.3, 10.3) (4.3, 8.4)
(6.0,10.1)
(6.7, 10.3) (7.7, 11.7) (11.7, 15.7) (11.2,15.3) (6.6, 10.7)
F-values df = (3, 1822) 59.2 39.4 18.6 39.4 38.0 65.7 68.9 21.6
'Each index is an equally weighted sum of the number of items indicated and is normalized to 100 points. Differences in weighted scorewIare all positive.
ples were analyzed for the urinary sodium and 24-hour dietary recall data. The home visits were scheduled by a mailed request followed by a phone contact. The 24-hour dietary recall by interview was selected because of the difficulty in obtaining multiple food records from young children. The 24-hour recall method has been shown to reliably measure group mean estimations of food intake in children.27 Interviewers were trained and certified to conduct the 24-hour food recalls with children and parents by the Nutrition Coding Center at the University of Minnesota. Interviewers explained the procedures for obtaining the overnight urine sample to the child and parent and distributed and collected the containers. At posttest only, those families who took part in the 24-hour dietary recalls also participated in an observation survey for targeted food items in their cupboards. This survey provided another objective measure of the food items in the child's environment. Development and reliability tests of this measure are detailed elsewhere.28 Eligibility requirements assured that all students selected for the home surveys were enrolled in one of the study schools during the intervention year 1985-86. Although all students participated in the school survey, only those who participated in the second and third grade surveys are included. Treatment assignment was determined by the school the student was enrolled in at the time of the posttest, allowing then for summer transfers. Analysis Data analyses were conducted separately for the school survey, 24-hour food recall, anthropometric, food shelf inventory, and urinary sodium data. For the anthropometric and school survey data, analysis of covariance (ANCOVA) was employed using the individual student as the unit of analysis and repeated, with schools nested within condition, to examine posttest scores for each index by condition, and with pretest scores on each index as a covariate. Confidence intervals were calculated using 1.96 times the standard error of the control posttest means (or the treatment minus control difference of means), using the pooled individual standard deviation obtained from the analysis of covariance. For the 24-hour recall and urinary sodium data, a mixed model two-way and three-way analysis of variance test was conducted to test for an overall effect, again first using the individual and then the school as the unit of analysis. The time by treatment interaction terms was explored, using tests of differences between group means. Confidence intervals were calculated using 1.96 times the standard error of the control posttest minus pretest means (or the treatment minus 1158
control differences), using the pooled individual standard deviation obtained from the analysis of variance. Analysis of variance was employed for the encouraged and discouraged foods in the food shelf inventory, examining Home Team (HT, HT/HH) versus no Home Team (HH, C) differences. All of these analysis methods are appropriate for unequal sample sizes of schools and students. Results Participation rates were notably high for all aspects of the study. For the Hearty Heart program, classroom observations by staff were undertaken, and full compliance to protocol was noted in all 16 participating schools. For the Home Team program, 86 per cent of the parents participated in at least one week's activities, and 71 per cent-nearly 1,000 families-completed the program. All of the students found it possible to find a suitable adult to participate with them in Home Team activities. Participation rates for the school survey and anthopometric measures were greater than 93 per cent for all measures at both pretest and posttest. Of the families who were contacted for the home survey, over 70 per cent participated at both pretest and posttest. The school survey measured knowledge, skills, and behavior as defined in the measures section of the questionnaire. Mean values for the control group, differences from the control, and the lower and upper bounds of the 95% confidence intervals at posttest for each index, adjusted for the pretest scores, are shown in Table 1 using the individual as the unit of analysis. The overall comparison (ANCOVA) between treatment conditions held up for all variables even when the school (N=31) was used as the unit of analysis (p
