Squires Quest!

Dietary Outcome Evaluation of a Multimedia Game

Tom Baranowski, PhD, Janice Baranowski, MPH, RD, LD, Karen W. Cullen, DrPH, RD, LD, Tara Marsh, MS, RD, Noemi Islam, MS, RD, Issa Zakeri, PhD, Lauren Honess-Morreale, MPH, Carl deMoor, PhD Background: Fruit, juice, and vegetable (FJV) consumption among children is low. Innovative programs are needed to enable children to increase FJV intake. Psychoeducational multimedia permits the delivery of interventions as designed and capitalizes on known behavior change principles. Design: Elementary school was the unit of recruitment, assignment, and analysis. Twenty-six elementary schools were pair matched on size and percentage of free or reduced-price lunch, and randomly assigned to treatment or control groups. Data were collected just before and just after the program.

Setting/ All fourth-grade students in participating elementary schools were invited to participate. Participants: Data were collected on 1578 students. Main Outcome: Servings of fruit, 100% juice, and vegetables consumed.

Intervention: Squires Quest! is a ten-session, psychoeducational, multimedia game delivered over 5 weeks, with each session lasting about 25 minutes. Based on social cognitive theory, educational activities attempted to increase preferences for FJV through multiple exposures and associating fun with their consumption, increase asking behaviors for FJV at home and while eating out, and increase skills in FJV preparation through making virtual recipes. Measures: Four days of dietary intake were assessed before and after the intervention. Assessment was made by the Food Intake Recording Software System (FIRSSt), which conducts a multiple pass, 24-hour dietary intake interview directly with the children. Children participating in Squires Quest! increased their FJV consumption by 1.0 servings more than the children not receiving the program.

Results:

Conclusions: Psychoeducational multimedia games have the potential to substantially change dietary behavior. More research is warranted. (Am J Prev Med 2003;24(1):52 61) 2003 American Journal of Preventive Medicine

Background
eople who consume more fruit, 100% juice, and vegetables (FJV) have greater longevity1 and some level of protection from several cancers,2 heart disease,3 diabetes mellitus,4 and perhaps even aging of skin.5 While children do not ordinarily experience adult chronic diseases, some cancers have a long developmental period, perhaps initiating at puberty.6
From the Childrens Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine (T Baranowski, J Baranowski, Cullen, Marsh, Islam, Zakeri), and M.D. Anderson Cancer Center, Department of Behavioral Science, University of Texas (HonessMorreale, deMoor), Houston, Texas Address correspondence and reprint requests to: Tom Baranowski, PhD, Professor of Pediatrics (Behavioral Nutrition), The Childrens Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine, 1100 Bates Street, Room 2038, Houston TX 77030-2600. E-mail: tbaranow@bcm.tmc.edu.

Furthermore, food-related preferences and practices start in the earliest years,7 and FJV consumption may track (i.e., those at higher levels of FJV consumption at younger ages remain higher consumers later in life).8 10 Thus, interventions to promote higher FJV consumption among children hold the promise of immediate healthier growth for children11; the prevention of the initiation of cancer during adolescence, if carried through puberty6; and the prevention of cancer and other illnesses in mid-life, if carried into adulthood. Nutrition education programs effect change in eating FJV through changing mediating variables.12,13 The most important mediators of eating more FJV include increasing the availability and accessibility of FJV at home and when eating out,14 increasing FJV preferences,7,15 and increasing childrens skills at making FJV

52

Am J Prev Med 2003;24(1) 0749-3797/03/$see front matter 2003 American Journal of Preventive Medicine Published by Elsevier Science Inc. PII S0749-3797(02)00570-6

recipes when they are responsible for making their own snacks.16 Schools present an important channel for reaching large numbers of children with nutrition education. Only one school nutrition education program, which was implemented by specially trained teachers, resulted in substantial FJV change (1.6 servings) after 2 years of program delivery.17 Others have achieved modest18,19 or no FJV change.20 One of these interventions revealed that the usual classroom teacher implemented only 50% of the curriculum-specied activities with only 22% of the activities likely to result in behavior change.21 Thus, channels need to be found that deliver nutrition education programs more directly to children. One channel that interacts with children directly is computer-based, interactive multimedia education (IMME).22 IMME is an attractive educational modality because it can combine visual, aural, and text-based messages22 and incorporate entertainment into education (edutainment), thereby making the messages more acceptable and the activities more enjoyable. When based on known psychological principles, IMME has been called psychoeducational multimedia training (PEMT).23 PEMT can deliver the intervention exactly as designed by the investigators, especially the precise specication of theoretically prescribed, behavior change procedures. Two early PEMT programs were targeted at high-risk behaviors among adolescents.24,25 More recent PEMT programs addressed physical activity,26 obesity,27,28 and eating disorders.29 A PEMT kiosk whose program was based on social cognitive theory in ve grocery stores led to improved consumption of dietary fat, ber, fruit, and vegetables.30 Adults receiving a talking computer intervention over the telephone increased fruit consumption by 1.1 servings per day.31 Computer-tailored, nutrition-education letters sent to families of adolescents resulted in reduced dietary fat intake32 among all family members; these letters worked, however, only with the mothers, not the fathers or adolescents.32 No research on PEMT to promote dietary change has been reported with elementary school children. This paper reports the dietary outcome evaluation of the Squires Quest! PEMT nutrition education game with fourth-grade students in Houston, Texas.

assessment took approximately 2 weeks per school, the intervention was conducted in 5 weeks, and post-assessment took approximately 2 weeks per school. Thus, a cycle could be completed in one school semester. Schools were matched on size33 and percentage of students receiving free and reducedprice lunch (an indicator of socioeconomic status [SES]). Within matched pairs, schools were randomly assigned to conditions. Power calculations revealed that 26 elementary schools were necessary to detect a 0.5 standard-deviation change in servings of FJV with 80% power and 5% type-one error assuming a school-associated intraclass correlation of 0.02. The design was implemented in two waves. Initially, 14 elementary schools were recruited and randomly assigned to groups in fall 1999. One control school declined to participate after baseline assessment. As a result, 13 elementary schools were recruited and randomly assigned to groups in spring of 2000. Due to the limited number of computers for baseline assessment and intervention (n 76), and the need to leave the computers in the treatment schools to run Squires Quest! (the variability in type and processing speed of computers already in the schools was too large to anticipate in the programming), the treatment and control schools were identied prior to baseline assessment. Baseline assessment was conducted for approximately 2 weeks in the control schools, and then computers were moved to the treatment schools for assessment and intervention. The treatment group completed post-assessment rst, after which the computers were moved to the control schools.

Sample
The Houston Independent School District (the third largest school district in the United States) agreed to participate. A comprehensive recruitment program was implemented, starting with contact with administrators and presentations to principals, and followed by presentations to teachers at schools with principals agreeing for their schools to participate. Decisions to participate were often made by the principal in consultation with the teachers. Details of the recruitment procedures and the factors inuencing school participation have been presented elsewhere.34 All children completed an informed assent form, and parents completed an informed consent form for their child and themselves, which was returned by the child to school. The consents were worded to indicate that the child could receive either condition, but measurement was required in both. In the treatment schools, 73.2% of students provided informed consent to participate, and 67.6% of students provided informed consent in the control schools. A total of 1578 students were recruited to participate in Squires Quest! Characteristics of the children in the 26 schools that completed post-assessments, as well as those who did not, appear in Table 1.

Methods Research Design

Our objective was to demonstrate dietary change immediately after implementation of the Squires Quest! program. We believed that it was important to demonstrate change with a new technology right after the intervention. To achieve this objective, a simple two-group design (treatment and control) with pre- and post-assessment was employed. Due to the effects of data clustering, the school was the unit of recruitment, random assignment to group, and analysis. Baseline

Squires Quest!
Squires Quest! was designed as a ten-session, interactive multimedia game, with each session taking about 25 minutes to complete. The story line for the game was as follows: the kingdom of 5A Lot was being invaded by the Slimes (snakes)

Am J Prev Med 2003;24(1)

53

Table 1. Sample characteristics and assessment of differences at baseline Completed pre- and post-assessment n (%) 1489 (94.7) 749 (50.3) 740 (49.7) 30 (2.3) 833 (63.3) 383 (29.1) 61 (4.6) 9 (0.7) 689 (47.5) 763 (52.5) 261 (17.9) 643 (44.1) 452 (31.0) 102 (7.0) Mean ( SD) 1.7 (2.1) 0.8 (1.5) 0.8 (1.2) 3.4 (3.4) 0.3 (0.6) 3.6 (3.5) Control Completed pre and post (n 749) 1.8 (2.2) 0.8 (1.6) 0.9 (1.3) 3.5 (3.7) 0.3 (0.7) 3.8 (3.8) Completed pre only (n 36) 1.8 (1.8) 0.6 (0.7) 0.7 (1.0) 3.0 (2.5) 0.4 (0.6) 3.4 (2.6) Completed pre-assessment alone n (%) 89 (5.3) 36 (40.4) 53 (59.6 ) 2 (2.7) 39 (52.0) 22 (29.3) 9 (12.0) 3 (4.0) 47 (54.0) 40 (46.0) 7 (8.6) 47 (58.0) 24 (29.6) 3 (3.7) Mean ( SD) 1.7 (2.0) 0.9 (2.3) 0.6 (0.9) 3.2 (3.6) 0.5 (0.9) 3.7 (3.7) Treatment group n (%) 749 (100) 749 (100) 26 (4.1) 413 (65.8) 159 (25.3) 28 (4.5) 2 (0.3) 344 (47.0) 373 (52.0) 140 (18.9) 326 (44.0) 229 (30.9) 46 (6.2) Mean ( SD) 1.8 (2.2) 0.8 (1.5) 0.8 (1.3) 3.5 (3.7) 0.3 (0.7) 3.8 (3.8) Completed pre and post (n 740) 1.6 (2.0) 0.8 (1.3) 0.8 (1.1) 3.2 (3.0) 0.3 (0.5) 3.5 (3.1) Control group n (%) 740 (100) 740 (100) 4 (0.6) 420 (61.0) 224 (32.6) 33 (4.8) 7 (1.0) 345 (46.9) 390 (53.1) 121 (16.9) 317 (44.2) 223 (31.1) 56 (7.8) Mean ( SD) 1.6 (2.0) 0.9 (1.5) 0.8 (1.1) 3.2 (3.1) 0.3 (0.5) 3.5 (3.2) Completed pre only (n 53) 1.7 (2.1) 1.1 (3.0) 0.6 (0.9) 3.4 (4.2) 0.5 (1.0) 3.9 (4.2)

Sample characteristics Totals Group Treatment Control Agea (in years) 8 9 10 11 12 Gender Boys Girls Ethnic groupb African American Euro-American Hispanic Other Dietary intakec (in servings) F J RV FJV HFV Total FJV HFV

Total sample N/n (%) 1578 (100) 785 793 32 872 405 70 12 736 803 268 690 476 105 Mean ( SD) 1.7 (2.1) 0.8 (1.5) 0.8 (1.2) 3.4 (3.4) 0.3 (0.6) 3.6 (3.5)

Intervention

Dietary intake F J RV Total FJV HFV Total FJV HFV

Note: n varies by variable due to missing data. a Chi-square 18.185, df 4, p 0.001 for pre- and post-assessment versus pre-assessment only comparison. Chi-square 27.733, df 4, p 0.001 for control versus intervention comparison. b Chi-square 8.362, df 3, p 0.039 for pre- and post-assessment versus pre-assessment only comparison. c t 2.18, df 1487, p 0.029 for control versus intervention fruit comparison, and t 2.69, df 1576, p 0.007 for control versus intervention HFV comparison. F, fruit; FJV, fruit, juice, and vegetables; HFV, high-fat vegetables; J, 100% juice; RV, regular vegetables.

and the Mogs (moles), who were attempting to destroy the kingdom by destroying the fruit and vegetable crops. King Cornwell and Queen Nutritia were leading their knights (e.g., Sir Sarah See-a-Solution and Sir Alex Try-to-be-Right) to defeat the invaders. In the rst session, the fourth-grade child committed to becoming a squire in the pursuit of becoming a knight to help the king and queen. The squire had to face challenges in his/her quest. The challenges involved skills and goals related to eating more fruit, 100% fruit juice, and vegetables. The squire prepared FJV recipes (in a virtual kitchen) to provide energy for the king and court to ght the invaders. A wizard mentored the child through the challenges, and the castle robot (Mad Maxie) facilitated many of the educational sessions. The invaders kidnapped the good chef (Chef Karat) and replaced him with Chef Mog, a

bumbling fool, who always usurped the squires accomplishments to make himself look good to the king. Before the end of each session, the child set goals to make the recipe (prepared in the virtual kitchen) during that session, eat another FJV serving at a meal or as a snack, or to ask for his/her favorite FJV to be more available at home. The children participated in a decision-making activity between their favorite fruit, juice, or vegetable and a more common snack. The FJV was selected based on the childs food preferences reported at baseline. The decision criteria were the three most important outcome expectancies reported by the child at baseline. Sessions 2 to 10 began with an assessment of whether the goal from the previous session was completed, for which dragon-scale points were assigned. A problem-solving routine

54

American Journal of Preventive Medicine, Volume 24, Number 1

Table 2. Squires Quest! game overview of sessions 1 through 10 Session 1 Day M-W Meal After-school snack Skill Decision making F vs non-F Knowledge What counts as F or J? What counts as J svg? How many svgs/day? What counts as F? What counts as F svg? Recipe J Drink 1 recipe F Drink 2 recipes 3 M-W School lunch Problem solving Decision making: V vs other snack 4 T-S Dinner V Side dish Selecting F/V Storage F/V Trash It game Stash It game 5 M-W Afternoon Problem solving specic to asking skill: asking with negotiating What foods count as V? What counts svg of V? How many svgs/day? How to buy fresh F? How to buy fresh V? How to store fresh? How to buy canned, frozen? How to substitute V? Substitution depends on time and preference Side Dish 3 recipes Snack V svg 2 days at snack Make and try recipe, ask ingredients Introduce 5-a-day plan Knight eats 2 at snack Week 3: parent newsletter Asking skill: Did you ask? (Yes/No) Yes: What happened? Got item Reinforcing message: Did not get item; list of most common problems with solution (no time, not available) No: Why not? (forgot, not comfortable) If Session 3 goal not done, no make-up
(continued on next page)

Goal setting/ assignment 1 svg of F/J 2 days at snack Make and try recipe, ask ingredients 1 svg of F/J 2 days at breakfast or snack Make and try recipe, ask ingredients Eat svg of V 2 days at school lunch 1 V svg 2 days at dinner Make and try recipe, ask ingredients

Schema None

Miscellaneous Week 1: parent newsletter

T-S

Breakfast

Asking skills

None

No Recipes

None

None

Why eat breakfast? Problem-solving routine if student does not complete goal session 1 Needs to be a routine that appears when goal has not been met Week 2: parent newsletter School lunch: what counts? If Session 1 goal not done, no make-up Use senses for guidelines If Session 2 goal not done, no make-up

Snacks

3 recipes

Am J Prev Med 2003;24(1) 55

56 Table 2. (continued) American Journal of Preventive Medicine, Volume 24, Number 1 Session 6 Day T-S Meal FF lunch or dinner Skill Decision making at FF restaurant Choosing F/V Knowledge What V choices at FF? What F choices at FF? How to make a F/V svg? What V choices at CS? What F choices at CS? What to look out for? Review F/V knowledge Why try new F/V? Taste buds change 9 M-W Desserts Decision making: F vs other dessert Review F/V goals met and how they can be related to developing schema Combo Foods 3 recipes Dessert Try eating 5-a-day schema Make and try recipe, ask ingredients Eat V svg at lunch and dinner for 2 days 5-a-day plan Wizard has 2 at dinner Student creates schema Recipe No Recipes Goal setting/ assignment 1 svg F/V at FF Restaurant alternatives for student; no FF Serving of F/V at CS Alternatives for no CS Schema 5-a-day plan King eats 2 F at breakfast 5-a-day plan Queen has 2 at lunch Week 4: parent newsletter Store will have J as replacement for soda If Session 5 goal not done, no make-up If Session 6 goal not done, no make-up Sweetness of F Week 5: parent newsletter If Session 7 goal not done, no make-up Outcome expectations: Good thing that happen when eat F/V Family Feudtype show: Survey says . . . Stronger, more energy, better eye sight If session 89 not done, no make-up Knighting ceremony! 3 recipes Present schema to king Goal: Choose 2 behavior changes from program to continue Miscellaneous If Session 4 goal not done, no make-up 7 M-W CS after school/ noon snack Find F/V at CS Find it and buy it game How ads manipulate 8 T-S V with lunch and dinner Knowledge FJ and V review game No Recipes 3 recipes 10 T-S Party Schema Fun F/V at parties Party foods Make goal to follow new schema to get 5-a-day everyday! Schema check; correct if needed Woven into recipe preparation
CS, convenience store; F, fruit; FF, fast food; J, 100% juice; svg, serving; V, vegetable.

was employed to help the child assess how he/she might change practices to increase the likelihood of goal attainment. Table 2 presents a list of session activities, in rough order of sequence within a session. All children attained knighthood (ten possible levels) at the end of the ten sessions, with the level determined by the number of dragon-scale points earned. Points were earned primarily by attainment of goals, with smaller amounts of points earned from the educational games. Examples of these games included identifying what counted as fruit, what counted as vegetables, and whether a demonstration of asking would likely result in making FJV more available. Focus group discussions with fourth-grade children were employed in the design of Squires Quest! to assess interest in the story line and to identify child-desired characteristics of characters.35 Squires Quest! focused on fourth-grade children alone, based on advice from developers of childrens games. The primary creative writer attended two national conferences of developers of interactive multimedia games to benet from their insights.

Table 3. Reliability intraclass correlations across 4 days of assessment at pre- and post-assessment Dietary intake Fruit 100% juice Vegetable Total fruit, juice, and vegetable Pre-assessment 0.44 0.38 0.38 0.59 Post-assessment 0.52 0.44 0.46 0.58

were added to the above in separate models. To control for non-normal distributions of the dietary variables, these analyses were repeated with log-transformed data. The sample was divided into quartiles of fruit (F); 100% juice (J); regular vegetables (RV); total fruit, juice, and vegetables (TFJV); high-fat vegetables (HFV); and total fruit, juice, and vegetables, with high-fat vegetables (TFJV HFV) consumption at baseline. Within quartiles, mean intake was calculated at baseline and at post-assessment separately for treatment and control groups.

Measures
The primary outcome measure was servings of fruit, 100% juice, and vegetable consumption as assessed by the Food Intake Recording Software System (FIRSSt). FIRSSt was administered for 4 days at baseline and at post-assessment. We attempted to obtain 4 nonconsecutive days per child over a 2-week period, but in some cases we obtained consecutive days. FIRSSt is an interactive, multimedia dietary assessment program that simulates a multiple-pass, 24-hour dietary recall (24hDR),36 completed by the child. FIRSSt was demonstrated to perform with 46% matches against observation of previous days school lunch and 60% matches against a dietitianconducted 24hDR.36 This was only somewhat lower than the dietitian-conducted 24hDR against the observation of previous days school lunch (59%).36 Demographic characteristics were assessed using standard questions on the consent form sent home to the parents.

Results
Only 5.3% of students did not complete pre- and post-assessments. The modal category for age was 9 years, with children completing pre- and post-assessment being slightly younger. Euro-American children were somewhat less likely than other groups to complete both pre- and post-assessments. There were no differences between participation groups in gender or servings of FJV consumed at baseline (Table 1). Despite random assignment of schools to condition, the children in the control group were slightly older, but there were no differences between groups by gender or ethnic group (Table 1). Children assigned to the treatment group consumed slightly more fruit and high-fat vegetables at baseline. At baseline, the intraclass correlation associated with the clustering effect by school was 0.02 for F, 0.01 for J, 0.02 for RV, and 0.03 for TFJV. The intraclass correlations for change from pre- to post-assessment were 0.03 for F, 0.04 for J, 0.01 for RV, 0.05 for TFJV, 0.02 for HFV, and 0.05 for TFJV HFV. The reliability-related intraclass correlations across 4 days of assessment were modest and comparable between pre- and post-assessments (Table 3). Children in the spring implementation wave were somewhat less likely to complete all ten sessions (Table 4). Although principals agreed to participate, we were asked to leave several schools early in the spring, when we came within 2 weeks of the administration of standardized tests, to enable teachers to concentrate on teaching to the standardized test. Using a mixed-model analysis, the difference in means between treatment and control groups at postassessment, after controlling for pre-assessment values, was 0.91 servings of FJV or 1.0 servings of TFJV HFV per day (Table 5). Contributions to this difference were
Am J Prev Med 2003;24(1) 57

Statistical Analyses
Differences in categoric variables between those completing pre- and post-assessments and those completing pre-assessment alone, and between treatment and control groups, were tested by chi-square statistics. Differences between these groups in mean servings of FJV were tested by independent t test. Analysis of variance was used to calculate intraclass correlations for the clustering effect of school and for reliability across 4 days. Independent sample t tests were used to test for differences between groups in consumption of FJV at baseline. Mixed-model analysis of covariance was used to test for differences between treatment and control groups (xed effect) in FJV and each FJV component at post-assessment, controlling for corresponding pre-intervention values and for the clustering effect of school (variable effect). Matching was not included in these analyses.3739 For methodologic rigor, these analyses were repeated using the mean for each school in a pair-matched t test, a weighted pair-matched t test, and a Wilcoxon-signed ranks test. To test for possible moderation of treatment effects by age, gender, or ethnicity, the demographic term and a group demographic interaction term

Table 4. Number of Squires Quest! sessions completed by students Number of sessions 1 2 3 4 5 6 7 8 9 10 Total Fall 1999 n (%) 10 (2.4%) 7 (1.7%) 3 (0.7%) 6 (1.4%) 4 (1.0%) 2 (0.5%) 5 (1.2%) 8 (1.9%) 10 (2.4%) 364 (86.9%) 419 (100%) Spring 2000 n (%) 1 (0.3%) 2 (0.5%) 2 (0.5%) 1 (0.3%) 7 (1.8%) 11 (2.9%) 19 (5.0%) 43 (11.2%) 71 (18.5%) 226 (59.0%) 383 (100%) Total n (%) 11 (1.4%) 9 (1.1%) 5 (0.6%) 7 (0.9%) 11 (1.4%) 13 (1.6%) 24 (3.0%) 51 (6.4%) 81 (10.1%) 590 (73.6%) 802 (100%)

made from each food group, but the statistically significant differences were for F (0.52 servings) and RV (0.24 servings) (Table 5). This analysis was repeated with a t test on the mean of each school weighted by number of students per school, a matched-pairs t test (not weighted), and a Wilcoxon-signed ranks test on matched school means. These analyses revealed similar results (Table 5). The analysis was also repeated deleting the one school that was not randomly assigned with the same results. There was no evidence for moderation of this effect by age, gender, or ethnicity. Girls ate more (main effect) fruit and vegetables, but not juice. Neither age nor ethnicity was related to any component of FJV consumption. The same pattern of ndings was found with log-transformed data, with the lone exception that the outcome effect for vegetables was reduced to marginal signicance (p 0.06). To describe the process of dietary change, FJV consumption is consumption group at baseline with mean values at baseline and post-program for treatment and control groups separately (Table 6). Substantial regres-

sion to the mean occurred in all food groups. In the third quartile, however, the mean increased for the treatment group and did not increase or declined in the control group.

Discussion
The Squires Quest! PEMT game resulted in a 1.0 serving difference of FJV between treatment and control groups at the end of the 5-week, ten-session program, after controlling for baseline FJV consumption. The strengths of this research include the large sample of schools and students, the mixed ethnic and SES composition of the sample, using the school as the unit of assignment and analysis, random assignment of school to condition, minimal differences in consumption by groups at baseline, the ability of the computer to deliver the intervention as designed, and similar results for all approaches to outcome analysis. It is not clear, however, how long this increased change would be maintained beyond the end of the program. The

Table 5. Differences in food group intakes between treatment and control groups at post-assessment from Squires Quest! using four analyses Mixed-models analysis of variance 0.52 F 9.47, p 0.17 F 2.02, p 0.24 F 10.6, p 0.91 F 9.4, p 0.09 F 2.6, p 1.01 F 11.7, p 0.002 0.156 0.001 0.002 0.107 0.0007 Weighted matched pairs t test (n1*n2)/(n1 n2) 0.70 F 7.64, 0.17 F 2.02, 0.27 F 8.36, 1.14 F 4.41, 0.12 F 8.52, 1.26 F 4.24, p 0.006 p 0.156 p 0.004 p 0.036 p 0.004 p 0.040 Matched pairs t test (not weighted) 0.66 t 3.458, p 0.19 t 1.761, p 0.24 t 2.70, p 1.09 t 3.63, p 0.11 t 2.11, p 1.20 t 4.127, p 0.005 0.104 0.019 0.003 0.057 0.001 Wilcoxon-signed ranks test on matched school means NA Z 2.76, NA Z 1.43, NA Z 2.27, p NA Z 2.83, NA Z 1.85, NA Z 2.90, p 0.006 p 0.15 0.023 p 0.005 p 0.064 p 0.004

Dietary intake Fruit 100% juice Regular vegetables Total fruit, juice, and vegetables High-fat vegetables Total fruit, juice, and vegetables with high-fat vegetables
NA, not applicable.

58

American Journal of Preventive Medicine, Volume 24, Number 1

Table 6. Mean baseline and post-assessment value for treatment and control by overall quartile at baseline Dietary intake Fruit I II III IV Juicea I, II III IV Regular vegetablesa,b I, II III IV Total fruit, juice, and vegetables I II III IV High-fat vegetablesa,b I, II, III IV Total fruit, juice, and vegetables I II III IV 0.07 (0.12) 0.71 (0.21) 2.57 (2.26) 0.04 (0.08) 0.69 (0.25) 2.49 (1.43) 0.50 (0.37) 1.73 (0.36) 3.31 (0.56) 7.87 (3.83) 791 399 388 777 412 389 399 390 393 396 0.57 (0.94) 0.90 (1.17) 1.71 (2.35) 0.60 (0.96) 1.04 (1.24) 1.76 (1.57) 1.93 (1.91) 2.88 (2.70) 3.96 (3.30) 6.21 (4.43) 0.35 (0.70) 0.59 (0.78) 2.48 (2.55) 3.08 (2.61) 4.65 (3.71) 6.59 (4.66) 360 204 176 349 212 179 171 191 202 176 574 165 170 200 203 167 0.57 (0.97) 0.70 (0.98) 1.26 (1.48) 0.48 (0.82) 0.70 (0.90) 1.37 (1.49) 1.52 (1.65) 2.11 (2.08) 3.16 (2.49) 4.81 (3.76) 0.27 (0.49) 0.66 (0.89) 1.76 (1.69) 2.59 (2.37) 3.35 (2.42) 5.24 (3.91) 384 173 192 378 182 188 202 181 168 198 590 159 203 172 174 200 0.06 (0.10) 0.68 (0.21) 1.56 (0.34) 4.53 (2.54) 396 394 399 389 1.06 (1.87) 1.31 (1.63) 1.85 (1.90) 3.22 (3.15) 184 196 197 163 0.54 (1.06) 0.79 (1.19) 1.17 (1.34) 2.72 (2.63) 188 175 183 203 Baseline quartiles Baseline M (SD) n Treatment M (SD) n M (SD) Control n

0.04 (0.10) 1231 1.18 (0.76) 347 with high-fat vegetables 0.61 (0.41) 397 1.98 (0.40) 392 3.67 (0.63) 397 8.36 (3.84) 392

Notes: Values removed were post-assessment regular vegetables (17.13), and post-assessment high-fat vegetables (15.00). New maximum values were post-assessment regular vegetables (7.75), and post-assessment high-fat vegetables (5.38). a At baseline, 25% of students had no juice intake, 25% of students had no regular vegetable intake, and 50% of students had no high-fat vegetable intake; therefore, it was necessary to combine some quartiles. b Removed one outlier ( 10 servings from all other values). M, mean; SD, standard deviation.

average consumption of FJV at baseline was comparable to that reported elsewhere in the literature among children.19,40 While the increase of 1.0 FJV servings in consumption was substantial, it was not enough to achieve the goal of an average of ve servings per day. This reinforces the idea that children need to be exposed to the ve-per-day message from multiple channels, and the messages likely need to be repeated several times throughout childhood in developmentally appropriate ways. The change in FJV consumption in this study is second largest in the literature after a serving difference of 1.6 for the High 5 Alabama program.17 The hallmark of both programs is that they were not implemented by the usual classroom teacher, who was documented to implement only about half the activities in corresponding curricula.41 The substantial changes attained suggest that the social-cognitive theory framework on which both interventions were predicated represent important aspects of behaviors and, thereby, provide a strong foundation for the design of dietary change programs for children. PEMT programs based on other theories would be useful to assess the extent to which the theory or medium accounts for outcomes.

The pattern of change from pre- to post-assessment by quartile (Table 6) was similar to other interventions.19 There was less of a decline in the highest quartiles of the treatment versus control groups and more of an increase in the lowest quartiles. These substantial changes by quartile occurred despite 4 days of pre- and post-assessment that should provide more stable estimates of intake. Thus, while a shift of about 1.0 servings occurred in the mean, this was not simply the addition of a serving to all cases. This pattern suggests that messages should be tailored based on baseline consumption: minimize decline in the higher groups and try it in the lower groups (where 25% of the sample were consuming no 100% juice or vegetables). This pattern likely explains the number of goals baseline consumptioninteraction effect reported elsewhere.42 The results reveal that psychoeducational multimedia games can result in dietary behavior change. It appears that the fun aspect kept the attention of the students and may have facilitated the change.43 It is not clear to what extent, or how, each of the following contributed to outcome: an interesting story line about aiding a king to fend off invaders, the interactivity of
Am J Prev Med 2003;24(1) 59

recipe preparation in the virtual kitchen, the tailoring of goal setting to baseline dietary assessment, the tailoring of decision making to baseline reports of FJV preferences and outcome expectancies, or other aspects of Squires Quest! Further tests of this technology should systematically vary these components to elucidate how this new technology inuences mediating variables and behavioral outcomes. Health educators can benet from more contact with developers who create games for children and by integrating theoretical behavioral frameworks into educational games. Creating such educational games is very expensive, requiring large teams of educational, dietary, and behavioral professionals, with subcontracting for professional artists and programmers. Periodically updating the program, based on feedback from children and teachers, could enhance the program but would add to cost. Alternatively, the possible conversion of this educational technology to the Internet holds out the promise of reaching large numbers of individuals, thereby minimizing marginal cost per new participant. Some technologic challenges need to be overcome (e.g., the lengthy download time that discourages participation) before a smooth transition from CD-ROM to the Internet can be made. Future uses of Squires Quest! could include renement and continued use in the classroom or conversion to an individual sequential game that does not have dened sessions on CD-ROM or the Internet. The reliability-related intraclass correlations across 4 days of consumption were modest overall, but higher for FJV combined. Reliability in dietary assessment is, in large part, a function of the number of days of assessment.44 It would be very challenging to collect 4 days of data from large groups of children in school settings using computerized procedures. Other limitations of this study include the limited completion of all ten sessions in the spring implementation (but suggests even larger changes may have been attained if higher participation were achieved) and the dietary assessment that relied solely on self-report. Squires Quest! demonstrates that PEMT games can induce dietary behavior change among elementary school children. Further research is warranted to identify key components in, and the duration of, effectiveness, as well as the generalizability of such games.
We are grateful to Colin McKay of SMILEX; Bruce Blausen of BMC Software, and Steve Hite, Mukesh Taylor, and Tom Robinson of Think Software, for their software programming; Electric Paintbrush and Sasha Fernandez, private entrepreneur, for their artwork; Brenda Congden, for development of creative content; Linda Zelley, MS, RD, for development of nutrition content; Felica Bradford, for data assessment; and Kathy Watson MS, MPH, for data analysis. This research was funded largely by the National Institutes of Health, grant R01 CA-75614. This work is also a publication

of the U.S. Department of Agriculture (USDA/ARS) Childrens Nutrition Research Center, at Department of Pediatrics, Baylor College of Medicine, Houston, Texas, funded in part by the USDA/ARS (Cooperative Agreement 58-62506001). The contents of this publication do not necessarily reect the views or policies of the USDA, nor does mention of trade names or organizations imply endorsement by the U.S. government.

References
1. Fraser GE, Shavlik DJ. Ten years of life: is it a matter of choice? Arch Intern Med 2001;161:164552. 2. Potter JD, ed. Food, nutrition and the prevention of cancer: a global perspective. Washington DC: World Cancer Research Fund and American Institute for Cancer Research, 1997. 3. Joshipura KJ, Hu FB, Manson JE, et al. The effect of fruit and vegetable intake on risk for coronary heart disease. Ann Intern Med 2001;134:1106 14. 4. Tuomilehto J, Lindstrom J, Eriksson JG, et al. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med 2001;344:134350. 5. Purba MB, Kouris-Blazos A, Wattanapenpaiboon N, et al. Skin wrinkling: can food make a difference? J Am Coll Nutr 2001;20:7180. 6. Colditz G, Frazier AL. Models of breast cancer show that risk is set by events of early life: prevention efforts must shift focus. Cancer Epidemiol Biomarkers Prev 1995;4:56771. 7. Birch LL. Psychological inuences on the childhood diet. J Nutr 1998; 128(suppl 2):40710. 8. Resnicow K, Smith M, Baranowski T, Baranowski J, Vaughan R, Davis M. Two-year tracking of childrens fruit and vegetable intake. J Am Diet Assoc 1998;98:7859. 9. Nicklas TA, Webber LS, Johnson CC, Srinivasan SR, Berenson GS. Foundations for health promotion with youth: a review of observations from the Bogalusa Heart Study. J Health Educ 1995;26(suppl):18 26. 10. Kelder SD, Perry CL, Klepp KI, Lytle LL. Longitudinal tracking of adolescent smoking, physical activity, and food choice behaviors. Am J Public Health 1994;84:11216. 11. Baranowski T, Mendlein J, Resnicow K, Frank E, Cullen KW, Baranowski J. Physical Activity and Nutrition (PAN) in children and youth: An overview of obesity prevention. Prev Med 2000;31(suppl):S1S10. 12. Baranowski T, Lin LS, Wetter DW, Resnicow K, Hearn MD. Theory as mediating variables: Why arent community interventions working as desired? Ann Epidemiol 1997;7:89 95. 13. Baranowski T, Anderson C, Carmack C. Mediating variable framework in physical activity interventions. How are we doing? How might we do better? Am J Prev Med 1998;15:266 97. 14. Hearn M, Baranowski T, Baranowski J, et al. Environmental inuences on dietary behavior among children: availability and accessibility of fruits and vegetables enable consumption. J Health Educ 1998;29:26 32. 15. Resnicow K, Hearn MD, Smith M, et al. Psychosocial correlates of fruit and vegetable consumption. Health Psychol 1997;16:2726. 16. Kirby S, Baranowski T, Reynolds K, Taylor G, Binkley D. Childrens fruit and vegetable intake: socioeconomic, adult, child, regional, and urbanrural inuences. J Nutr Educ 1995;27:26171. 17. Reynolds KD, Franklin FA, Binkley D, et al. Increasing the fruit and vegetable consumption of 4th graders: results from the High 5 project. Prev Med 2000;30:309 19. 18. Luepker RV, Perry CL, McKinlay SM, et al. Outcomes of a eld trial to improve childrens dietary patterns and physical activity: the Child and Adolescent Trial for Cardiovascular Health (CATCH). The CATCH collaborative group. JAMA 1996;275:768 76. 19. Baranowski T, Davis Hearn M, Resnicow K, et al. Gimme 5 fruit, juice and vegetables for fun and health: outcome evaluation. Health Educ Behav 2000;27:96 111. 20. Resnicow K, Davis M, Smith M, et al. Results of the Teach Well Worksite Wellness Program. Am J Public Health 1998;88:250 7. 21. Davis M, Baranowski T, Resnicow K, et al. Gimme 5 fruit and vegetables for fun and health: process evaluation. Health Educ Behav 2000;27:16776. 22. Morris JM, Owen GS, Fraser MD. Practical issues in multimedia user interface design for computer-based instruction. In: Reisman S, ed. Multi-

60

American Journal of Preventive Medicine, Volume 24, Number 1

23. 24.

25.

26.

27.

28. 29.

30.

31.

32. 33.

media computing, preparing for the 21st century. Harrisburg, PA: Idea Group Publishing, 1996:65105. Shavinina LV. Interdisciplinary innovation: psychoeducational multi media technologies. New Ideas Psychol 1998;16:189 204. Paperny DM, Aono JY, Lehman RM, Hammar SL, Risser J. Computerassisted detection and intervention in adolescent high-risk health behaviors. J Pediatr 1990;116:456 62. Bosworth K, Bustafson D, Hawkins RP, BARN Research Group. The BARN system: use and impact of adolescent health promotion via computers. Comput Human Behav 1994;10:46782. Prochaska JJ, Zabinski MF, Calfas KJ, Sallis JF, Patrick K. PACE , interactive communication technology for behavior change in clinical settings. Am J Prev Med 2000;19:12731. Turnin MC, Bourgeois O, Cathelineau G, et al. Multicenter randomized evaluation of a nutritional education software in obese patients. Diabetes Metab 2001;27:139 47. Tate DF, Wing RR, Winett RA. Using Internet technology to deliver a behavioral weight loss program. JAMA 2001;285:11727. Winzelberg AJ, Eppstein D, Eldredge KL, et al. Effectiveness of an internet-based program for reducing risk factors for eating disorders. J Consult Clin Psychol 2000;68:346 50. Anderson ES, Winett RA, Wojcik JR, Winett SG, Bowden T. A computerized social cognitive intervention for nutrition behavior: direct and mediated effects on fat, ber, fruits, and vegetables, self-efcacy, and outcome expectations among food shoppers. Ann Behav Med 2001;23:88 100. Delichatsios HK, Friedman RH, Glanz K, et al. Randomized trial of a talking computer to improve adults eating habits. Am J Health Promot 2001;15:21524. de Bourdeaudhuij I, Brug J. Tailoring dietary feedback to reduce fat intake: an intervention at the family level. Health Educ Res 2000;15:449 62. Barker R, Gump P. Big school, small school. Stanford, CA: Stanford University Press, 1964.

34. Currie CS, Elton RA, Todd J, Platt S. Indicators of socioeconomic status for adolescents: the WHO health behaviors in school aged children survey. Health Educ Res 1997;12:38597. 35. Cullen K, Baranowski T, Baranowski J. Computer software design for childrens food intake recording: What do third, fourth, and fth grade children like? J Nutr Educ 1998;30:4059. 36. Baranowski T, Islam N, Baranowski J, et al. The Food Intake Recording Software System (FIRSSt) is valid among 4th grade children. J Am Dietet Assoc 2002;102:380 5. 37. Martin DC, Diehr PH, Perrin EB, Koepsell TD. The effect of matching on the power of randomized community interventions studies. Stat Med 1993;12:329 33. 38. Diehr P, Martin DC, Koepsell TD, Cheadle A. Breaking the matches in a paired t-test for community interventions when the number of pairs is small. Stat Med 1995;14:1491504. 39. Proschan MA. On the distribution of the unpaired t-statistic with paired data. Stat Med 1996;15:1059 63. 40. Domel S, Baranowski T, Davis H, et al. Development and evaluation of a school intervention to increase fruit and vegetable consumption among 4th and 5th grade students. J Nutr Educ 1993;25:3459. 41. Hearn MD, Baranowski T, Baranowski J, et al. Environmental inuences on dietary behavior among children: availability and accessibility of fruit and vegetables enable consumption. J Health Educ 1998;29:26 32. 42. Cullen KW, Zakeri I, Pryor EW, Baranowski T, Baranowski J. Goal setting is differentially related to change in fruit and vegetable consumption among 4th grade children. Health Psychol In press. 43. Borra ST, Schwartz NE, Spain CG, Natchipolsky MM. Food, physical activity, and fun: inspiring Americas kids to more healthful lifestyles. J Am Diet Assoc 1995;95:816 23. 44. Baranowski T, Baranowski J, Doyle C, et al. Estimation of servings of fruit and vegetables and fat practices from adults seven day food records. J Nutr Educ 1997;29:3216.

Am J Prev Med 2003;24(1)

61