Main

Good nutrition and active life style are two building blocks for a healthy life (1). Children in big cities including Hong Kong (HK) have less exposure to nature (2,3), engage in sedentary activities, have poor eating habits, and get insufficient sleep compared with the previous generations (4,5), hence children’s health is becoming a prominent public concern, resulting in a rise in obesity prevalence (6); 4.7% of the HK 2 y olds are overweight/obese (4,5,6). At the same time, HK parents employ inappropriate strategies in managing the physical activities and dietary habits of their children, such as forced feeding, watching TV while eating, etc (4,5,6). Our intervention therefore targeted parents of young children as well as foreign domestic workers (domestic helpers), as they not only perform household chores but are highly involved in taking daily care of the children in HK (7,8), influencing their habits and routines (9). Nature-related activities enhance general well-being, as reflected in increased physical activity, a healthier diet, reduced stress, and better sleep (10,11,12,13). Most of the available intervention programs focus almost exclusively on diet and physical activity (14), and although some health promoting programs (15,16,17) worldwide have started to include gardening and connections to farms, none in HK to the best of our knowledge has taken into consideration a family’s interactions with nature. Health habits are developed very early in life, making it critical to start lifestyle interventions as early as possible (18). Hence, it is timely and highly relevant to develop new multi-dimensional lifestyle intervention programs and test their effectiveness for preschool children.

The main aim of the pilot study was to develop and test the study protocols, session scripts, logistics, measuring instruments, and program adherence. The secondary aim was to test the hypothesis that the intervention activities may induce positive changes in caregivers’ feeding habits, their physical activity levels, and eating and activity habits in preschool children.

Methods

Conceptual Influences and Theoretical Framework

Similar to INFANT (19), one of the most effective lifestyle parenting programs, the Play&Grow intervention used anticipatory guidance framework (20,21) in addressing the caregivers’ feeding, eating, and sedentary habits of children aged 2 to 4 (22). The detailed protocol of “Play&Grow” is described elsewhere (23).

Participants

Children from 2 to 4 y of age and their caregivers in HK were recruited and allocated to one of the two study sites, according to their residential addresses. Both parents were encouraged to participate, but largely only the domestic helpers (all females) and sometimes the mothers attended the sessions. The same caregivers were required to complete the questionnaires and to attend at least 80% of the program. After three families dropped out, 39 preschoolers (21 boys and 18 girls), 38 mothers and 38 domestic helpers (all females) chose to participate in the program (one family had two children). Three families, who decided to withdraw from the program, due to the lack of time did not differ in demographic characteristics from the rest of the families. The study was approved by the University of Hong Kong Human Research Ethics Committee for ethical clearance for research involving human participants (nr: EA1502073). It was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans.

The demographic data of the caregivers and children were collected via questionnaires. The anthropometric data (height, weight) was aggregated according to the international age-adjusted references (24,25) and converted to BMI (kg/m2) for each adult participant, and standardized for age and gender for children using BMI-for-age z-scores (26,27). The sample quality control performed during data entry, a second research assistant re-entered the data with randomly chosen sample questionnaires and found over 90% of consistency between these coders. All of the demographic and anthropometry data are presented in Table 1 . No significant difference in any anthropometry data was detected between the groups of the two sites (New Territories and Hong Kong Island). Mean age of the preschoolers was 33.97 mo, with a mean z-BMI of −9.38. The domestic helpers (n = 28, 73.7%) were responsible for cooking meals for children, and took the main responsibility of looking after children (n = 20, 52.6%).

Table 1 Participant characteristics at baseline measurement
Full size table

Intervention

Table 2 provides a brief description of each of 12 discussion sessions and parenting practices. Educational strategies included direct instruction, caregiver peer supports, group discussions and homework tasks. All the elements were developed prior to the start of the intervention. Each session, led by a group leader and a co-leader comprised of: (i) 15 min of guided active play involving both children and caregivers; (ii) 15 min of interactive education and skill development for caregivers; simultaneously supervised active play with foods for children, to promote acceptance of vegetables, and (iii) 15 min of guided active nature games outdoors, involving both children and caregivers. Each session included a lifestyle component (regarding eating, physical activity/sedentary behaviors and “connectedness to nature”) aimed at providing caregivers with knowledge and skills to support their child’s healthy lifestyle. To compensate for limited “theory time”, some additional “homework” was given to the families and feedback was provided during the next session. Examples of the additional novel “connectedness to nature” elements were: (i) 1/3 of each session included active outdoor nature games, focusing on developing motor skills, discovering nature, and practicing awareness of sounds, touch, smells, and temperature; (ii) group activities, practicing nature tasks; (iii) nature homework with follow-ups (collecting nature subjects and making art, growing own plants); (iv) environmental care training (reducing paper waste, water saving, recycling); (v) studying new vegetables and cooking together with them (avocado, avocado boat, guacamole, etc.).

Table 2 Intervention content
Full size table

Outcomes and Their Assessments

Certain family-related factors were expected to be associated with the effectiveness of the intervention program. The caregivers after completing the program were expected to: (i) know more about healthy feeding, eating, active play and (ii) gain general and specific parenting practices, aiming to establish healthy habits. The children would: (i) eat healthier, be less fussy with food; (ii) be more active, less sedentary, and (iii) have a higher “connectedness to nature”, reflected in increased time spent outside. The primary outcomes were categorized into feeding and eating habits; physical activity of caregivers and children’s active play and “connectedness to nature” habits were deemed as secondary outcomes. Since no similar interventions in this age group in HK had taken place, we adopted the MEND assessment design (28). To test the effectiveness of this 4-mo pilot study, the tests on lifestyle habits were administered prior to and immediately after the intervention period. Figure 1 presents the general enrolment flow chart of this pilot study. Examples of some items used in the questionnaires are provided in Table 3 .

Figure 1
figure 1

Enrolment flow chart.

PowerPoint slide

Full size image
Table 3 Examples of some items used in the questionnaires for Play&Grow
Full size table

Dietary Habits, Physical Activity Habits, and Connectedness to Nature

The Parental Feeding Style Questionnaire (PCFQ) (29) was used to assess parents’ feeding practice by four subscales (instrumental feeding, emotional feeding, promoting and encouragement to eat, and control over eating) with 27 items. Children’s eating behaviors were assessed using a 45 items HK Children’s Dietary Habit Questionnaire (HKCDHQ, being currently validated) (30). The items of PCFQ and HKCDHQ were rated on a five-point Likert-scales, from 1 point (never) to 5 points (always). Physical activity duration was measured in minutes (including light-, moderate-, and vigorous intensity) and sedentary behavior was assessed via the International Physical Activity Questionnaire (31) for caregivers and Physical Activity Questionnaire for Preschool-aged Children (Pre-PAQ) (32,33) for children. “Connectedness to nature” in children was measured using a nature relatedness scale, which is currently being validated by our research team. The intervention sessions ran once a week, and the intervention adherence was measured by the number of dropouts and qualitative group interviews at the beginning and the end of the study. Program orientation and debriefing meetings were held with program participants to provide some useful contextual information with regard to program participants’ expectations and experience of “Play&Grow” activities. As such, several discerning quotes are interspersed in the Discussion sections.

Statistical Analysis

All the data analyses were conducted using SPSS 20 (SPSS, Chicago, IL). Mean and SD were used to present continuous descriptive variables, frequency (percentage) for categorical variables. Due to the small sample size (n = 38), Wilcoxon signed-rank test or sign test was used to compare pre- and postintervention differences. The r code indicated the effect size. Correlation analysis was performed between the family-related factors and the effectiveness of the intervention. Spearman’s rank correlation was used for continuous variables, while Kruskal-Wallis test was used for categorical variables. Variables selected from correlation tests were included in a simultaneous multiple regression model. Results were considered as statistically significant when P < 0.05, two-tailed.

Results

Instrumental feeding (IF) and promoting and encouragement to eat (PE) behaviors of the caregivers improved significantly after the intervention. Among the 38 families, the caregivers’ IF behaviors reduced significantly after the intervention (z = −2.416, P = 0.016, r = 0.391). In addition, after participating in the Play&Grow program, caregivers performed positive parenting practices (i.e., promoting and encouragement to eat behaviors) more often (z = −2.652, P = 0.008, r = 0.430). Furthermore, the data derived from the HK Children’s Dietary Habit Questionnaire showed a significant improvement in the “battle during mealtime” (exact P = 0.007), “picky eating” (exact P = 0.049), “allowing consumption of salty food” (exact P = 0.039), and “supervision during the meals” (z = −2.285, P = 0.022, r = 0.371).

Domestic helper’s vigorous (z = −2.099, P = 0.036, r = 0.341), moderate (z = −2.127, P = 0.033, r = 0.345) active time and mothers’ vigorous active time during weekdays increased significantly during the intervention (exact P = 0.015). However, no differences were detected on child’s playtime (as reported in the questionnaires). Table 4 presents all of the intervention outcomes.

Table 4 Comparison between pre- and post-test
Full size table

Factors That Influenced the Effectiveness of Intervention

Families with a mother being responsible for child’s meal (x2 = 44.952, P = 0.026) and cooking (x2 = 4.656, P = 0.031), child’s birth weight (rs = −0.433, P = 0.007), and mother’s BMI (rs = 0.362, P = 0.028) were found to have a moderate bearing on the improvement of PE habits as one of a programs outcome. The domestic helper responsibility for child’s cooking (x2 = 7.342, P = 0.007), meals (x2 = 3.878, P = 0.049), and her instrumental feeding practices (rs = 0.393, P = 0.015.) correlated with improvement in child’s picky eating behavior. The child’s consumption of salty food correlated with mother’s responsibility for the child (x2 = 4.215, P = 0.040), and helper’s responsibility for cooking (x2 = 4.075, P = 0.044). Lastly, the supervision of the child during the meal correlated with the child’s z-BMI (rs = 0.369, P = 0.025), child’s age (rs = −0.392, P = 0.015), and mother’s age (rs = −0.327, P = 0.048). Table 5 presents all the potential factors, which might influence participants’ dietary habits. The influence on domestic helper’s moderate active time correlated with the numbers of computers at home, x2 = 6.842, P = 0.009. Table 6 presents all the potential factors, which might influence participants’ active time. None of the other family-related factors correlated with the effectiveness of intervention.

Table 5 Screening for factors which influenced the effectiveness on dietary habits
Full size table
Table 6 Screening for family–related factors which influenced the effectiveness on physical active time
Full size table

First, only three of the previously mentioned potential factors (i.e., mother’s BMI, child’s birth weight, the responsibility for child’s meal) together significantly contributed to predicting the influence on PE behavior in the simultaneous regression model. These three factors could together predict the outcome of this program on caregiver’s PE behavior, F (3, 33) = 3.540, P = 0.025, r2 = 0.243 and could explain 24.3% of the variability of the effectiveness of intervention on caregiver’s PE behavior. Second, only two previously mentioned potential factors (i.e., IF and responsibility for cooking) together could significantly contribute to a regression model and predict the effectiveness of this program on the “picky eater” behavior, F (2, 34) = 4.322, P = 0.021, r2 = 0.203, explaining 20.3% of the variability of the effectiveness. Lastly, the two factors (supervision of the child during the meals and responsibility for cooking the child’s meal) could predict the outcome of “allowing the child to consume salty food”, F (2, 34) = 5.737, P = 0.007, r2 = 0.252; this could explain 25.2% of the variability of the effectiveness of intervention regarding this behavior. All of the regression results can be found in Table 7 .

Table 7 Summary of the simultaneous multiple regression analysis of univariate predictors for the influences on dietary outcomes
Full size table

Discussion

A recent Cochrane review reported only a marginally significant improvement on body weight after some interventions and little to no effect on dietary or physical activity behaviors (34). It is worth noting that, in contrast to the study reported in the Cochrane review, our intervention program, even in the pilot form, indeed found highest impact on the caregivers feeding habits: the instrumental feeding (IF) and promoting and encouragement to eat (PE) behaviors of the caregivers improved significantly after the intervention. This finding is of relevance as parental dietary behavior is a significant predictor for child’s dietary habits (35). Several family-based intervention programs have demonstrated efficient outcomes on preschoolers (19,36) and prepubescent children (37) lifestyles. There is indication that the acquisition of healthy dietary habits (such as adequate vegetable/fruit consumption and dairy-rich intake) of young children participating in health promotion interventions can contribute to sustainable benefits after 2 y (34).

As suggested above, the parents have profound influence on children’s food habits and preferences (35). Our findings underline the importance of acknowledging and providing training for all caregivers in the family setting, including domestic helpers who influence children’s eating habits and physical activity patterns. We found that when the domestic helper was responsible for child’s cooking, her instrumental feeding practices predicted improvements in child’s picky eating behavior. Despite the mother being a primary caregiver, if the helper took responsibility for the cooking, this would lead to healthier eating habits of children, as for example, decrease in the consumption of salty foods after the intervention. It is quite possible that domestic helpers play a more important role in influencing the development of food acceptance patterns of young children than previously anticipated as the domestic helpers spend much time with children and become their natural role models (7) in HK (8). The intent of the program was to include the domestic helpers as well as the parents and intensively engage both in the program. Comments made by several parents and caregivers during the post-program debriefing sessions provide support for this goal being achieved: “…I want to highlight that as a family, we had a “stranger,” a “helper.” “She is now a part of the family, and through this program I feel our visions become uniform. Now she knows what we want, how she wanted to guide my daughter.”

In general, most of the family-based interventions promoting healthy lifestyle behaviors target parents or grandparents. To our knowledge, none of them has taken into account the common practice in HK, where domestic helpers are highly involved in the upbringing of children, creating a so called “extended family”. Therefore, the foreign domestic workers who have caregiver roles in the families should be actively involved, together with the other significant caregivers, in the delivery of healthy lifestyle interventions for young children.

As we found in our study, the primary reason why the families did not sign up for the program or left the program before completion was time constraints. This is consistent with the findings of other obesity intervention program research that points to how the time crunch experienced by many families presents a major obstacle to their participation in needed health education/lifestyle modification type programs (38). To address this issue, the viability of offering the program in different formats may be considered, e.g., on weekends instead of weekdays, for an intensive 3–4 d program instead of spread out over 4 months, and perhaps even via online delivery.

One of the aims of this pilot program was to evaluate the “connectedness to nature” element, as it is crucial to understand whether nature based health-promoting interventions can be effective. Several relevant tools are available, but all of them target adults and older children, e.g., the “Children’s connection to nature index” (38) and were not suitable for HK children of much younger age. We therefore further adapted the scale, creating a robust and valid instrument, suitable for studies associated to nature relatedness for preschool children in HK (under review).

To some extent, the lack of program impact on participating children playtime runs counter to findings from research indicating that programs with nature-related activities tend to increase physical activity (10,11,12,13). Hence, further possibilities were found for amplifying child’s exposure to natural elements, such as conducting outdoor activities on open grassy areas (as recommended in a US Institute of Medicine report on prevention policies to address early childhood obesity) (16), including more activities that weave in plants and gardening as ways to teach about nutrition (15) and reducing reluctance to try new foods including fruits and vegetables (17). Additional ideas for expanding on the platform of “connectedness to nature” related activities come from the families themselves. Some parents mentioned an interest in participating in multi-family play groups engaged in nature-related activities; however, they felt the need of help in constructing and conducting such activities. Several caregivers illustrated how their children gained a better understanding of the connection between food and nature: “My son learned the food that we eat is coming from the ground and he appreciates the farmer and the land.” “We used to ignore the growing process of foods, which indeed involved efforts from farmers, as well as nurtured by the nature. My kid has learnt that natural ingredients, water and soil are important in terms of “producing” foods, instead of buying food from market.”

To gain a better insight into how families communicate, cooperate, and make plans and decisions with regard to eating habits, physical activity and ways of engaging nature, a more extensive and formal qualitative method would be necessary to complement the quantitative data. One strategy would be to conduct an extensive qualitative interview process, possibly utilizing a focus group format similar to the one developed to explore family communication dynamics related to eating habits (39).

Drawbacks and Limitations

Some limitations of this study should be noted, keeping in mind that this study was a useful resource to test and adjust the methods, instruments, and logistics. One of the limitations was that results from the small sample size may not be representative for the entire HK population. Furthermore, the pilot identified some drawbacks in measuring instruments (e.g., subjective measurements of physical activity and child’s eating behaviors). Physical activity levels and times data was obtained from a subjective instrument (International Physical Activity Questionnaire and Physical Activity Questionnaire for Preschool-aged Children (Pre-PAQ)), and in the future randomized controlled trial (RCT) we will consider including a more objective measurement, such as pedometers or ActiGraphs. Although some improvements were found in children’s eating behaviors (e.g., in less conflicts during mealtime and less “picky eating”), these results should be considered as tentative findings as the Hong Kong Children’s Dietary Habit Questionnaire (HKCDHQ) instrument is still under validation. Thus, before starting the RCT, we will complete the validation first or choose another validated questionnaire (Children’s Eating Behaviour questionnaire (40)). In addition, since the nature relatedness scale is currently being validated by our research team, we did not report those results in this pilot study.

The immediate effect after the program is encouraging, and a follow-up study to examine whether this intervention has a lasting effect when comparing to a control group of similar age without receiving a similar educational program is a natural next step. The following practices should be measured and added to the upcoming RCT: Parental self-efficacy for correct practices, Limit settings, Parental modeling of healthy habits, diet, active play and “connectedness to nature”.

The long-term intention of “Play&Grow” was to generate a final program with long-term sustainable health benefits to HK’s current and future population. Over time, these benefits may be reflected in reduced prevalence of obesity amongst children and a concomitant reduction in healthcare costs in the public and private sectors. Furthermore, the results of the study could be of relevance to people living in other metropolitan areas or cities, which extends the impact of the study from local to global communities. In addition to these primary impacts, the study has an added scientific value by developing several important protocols and measurement instruments, especially the Nature Relatedness Scale for preschool children (subsequent work).

Conclusion

To the best of our knowledge, this is the first study of its kind in HK, promoting healthy lifestyle early in life and targeting not only the parents but the domestic helpers as well. The results therefore are expected to add value to the existing recommendations on physical activity and diet in children and, by introducing an environmental factor, “connectedness to nature”, to the healthy lifestyle recommendations. The impact of some of the actions taken during this pilot project are recognized, acknowledged, and highly appreciated by the caregivers and especially the foreign domestic workers in participating families.

In conclusion, by conducting the “Play&Grow” pilot study, we have successfully evaluated study design, protocols, methodology, and logistics. The strategies were adopted and modified and are now deemed suitable for a larger-scale RCT to promote healthy lifestyles for young children and their families.

Author Contributions

T.S. is the principal investigator of the study, she initiated and laid the main design of the study. All the authors participated in further development of the design, tools and wrote the manuscript. All the authors read and approved the final manuscript.

Statement of Financial Support

The study is supported by HKU SEED funding, GHELCH and Knowledge Exchange Grant.

Disclosure

The authors declare no competing interests.