Hostname: page-component-7b7567c8c-fvjsw Total loading time: 0 Render date: 2026-11-12T04:53:59.359Z Has data issue: false hasContentIssue false
  • English
  • Français

Variation in outcomes of the Melbourne Infant, Feeding, Activity and Nutrition Trial (INFANT) according to maternal education and age 2 and 3·5 years post-intervention

Published online by Cambridge University Press:  11 January 2021

Christine Delisle Nyström,
Adrian J Cameron,
Karen J Campbell and
Kylie D Hesketh
Christine Delisle Nyström*
Affiliation:
Department of Biosciences and Nutrition, Karolinska Institutet, Neo, Huddinge 141 83, SwedenInstitute for Physical Activity and Nutrition (IPAN), Deakin University, Geelong, Australia
Adrian J Cameron
Affiliation:
Global Obesity Centre, Institute for Health Transformation, Deakin University, Geelong, Australia
Karen J Campbell
Affiliation:
Institute for Physical Activity and Nutrition (IPAN), Deakin University, Geelong, Australia
Kylie D Hesketh
Affiliation:
Institute for Physical Activity and Nutrition (IPAN), Deakin University, Geelong, Australia
*
*Corresponding author: Email christine.delisle.nystrom@ki.se
Rights & Permissions [Opens in a new window]

Abstract

Objective:

This study aimed to assess whether the long-term effectiveness of the Melbourne Infant, Feeding, Activity and Nutrition Trial (INFANT) at 2 and 3·5 years post-intervention varied according to maternal education and age.

Design:

Two and 3·5 years post-intervention follow-up of the INFANT cluster-randomised controlled trial. Outcomes at both follow-ups included children’s BMI z-scores, physical activity (ActiGraph), television viewing (parental report) and dietary intake (3 × 24-h dietary recalls). Dichotomous moderator variables included maternal education (university v. no university) and age (< 32 v. ≥ 32 years).

Setting:

Population based.

Participants:

Families completing the 15-month programme (n 492) were invited to participate in the follow-ups when their child was 3·6 and 5 years old.

Results:

At the 2-year follow-up, the intervention effects on vegetable (positive) and sweet snack (negative) intake were greater in children with higher educated mothers, whereas water consumption (positive) was greater in children with lower educated mothers. At the 2-year follow-up, the intervention was more effective in increasing water consumption in children with younger mothers and decreasing sweet snack intake in children with older mothers (opposite result observed at the 3·5-year follow-up). At the 3·5-year follow-up, children with younger and older mothers increased and decreased their consumption of savoury snacks, respectively.

Conclusions:

Moderation by maternal education and age were observed for some outcomes; however, clear patterns were not evident at both follow-ups, with little consistency across outcomes. This indicates that INFANT was more-or-less equally effective in children irrespective of their mother’s education level or age, which is important in community-based interventions.

Keywords

ChildrenLong-term follow-upModerationNutritionPhysical activity

Information

Type
Research paper
Information
Public Health Nutrition , Volume 24 , Issue 6 , April 2021, pp. 1460 - 1468
Check for updates
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
© The Author(s), 2021. Published by Cambridge University Press on behalf of The Nutrition Society

Childhood overweight and obesity are growing public health issues worldwide regardless of country income. Approximately 38 million children under 5 years are classified as overweight or obese globally(1). More specifically, recent Australian data show the prevalence of overweight and obesity is 11 and 9 %, respectively, among children aged 2–4 years(Reference Huse, Hettiarachchi and Gearon2). This is of concern as overweight and obesity track from childhood into adolescence, with a recent study finding that approximately 90 % of 3-year-olds with obesity were overweight or obese when they were between 15 and 18 years of age(Reference Geserick, Vogel and Gausche3). Furthermore, Geserick et al. (Reference Geserick, Vogel and Gausche3) found that for adolescents with obesity, the fastest weight gain happened between the ages of 2 and 6. Reversal of overweight and obesity through interventions is difficult once established, demonstrating the urgent need to focus on primary prevention(Reference Waters, de Silva-Sanigorski and Hall4).

The Melbourne Infant, Feeding, Activity and Nutrition Trial (INFANT) was a community-based cluster-randomised controlled trial (2008–2010) focusing on the prevention of obesity-related behaviours in first-time Australian parents with infants aged 4 months(Reference Campbell, Hesketh and Crawford5). Post-intervention (15 months after baseline), children in the intervention group compared to the control group consumed significantly less sweet snacks and watched significantly less television(Reference Campbell, Lioret and McNaughton6). At the long-term follow-ups at 2 and 3·5 years post-intervention, that is, when the children were 3·6 and 5 years old, sustained effects for some obesity-related behaviours were observed. More specifically, positive intervention effects were observed for fruit, vegetable, sweet snack and water intake at the 2-year follow-up and for non-core drinks and sweet snack intake at the 3·5-year follow-up. Additionally, at both time points television viewing was around 10 min less per day in children in the intervention group(Reference Hesketh, Salmon and McNaughton7).

Maternal education has been found to be a strong predictor of child BMI, with lower maternal education associated with higher BMI from as early as 3 years of age(Reference McCrory, Leahy and Ribeiro8). Furthermore, in a recent systematic review, it has been observed that higher maternal educational attainment and age (≥ 30 years) are related to healthier dietary patterns in infants(Reference Gutierrez-Camacho, Mendez-Sanchez and Klunder-Klunder9). As maternal education and age have been found to be important determinants in obesity-related behaviours in childhood, it is important to assess the variations in outcomes of interventions by these indicators. Following INFANT, it was found that maternal age and educational attainment moderated the intervention effects for some obesity-related behaviours in infants aged 20 months(Reference Cameron, Ball and Hesketh10), that is, additional positive impacts of the intervention were observed for these subgroups. More specifically, the intervention increased vegetable intake and decreased sweet snack consumption in infants whose mothers had a higher education, whereas the intervention significantly increased water consumption in mothers with a lower education. With regard to maternal age, it was found that the intervention increased the consumption of both vegetables and water in infants with younger mothers(Reference Cameron, Ball and Hesketh10).

There is strong support for long-term follow-up of intervention studies in order to examine sustained effects(Reference Waters, de Silva-Sanigorski and Hall4,Reference Brown, Moore and Hooper11) . However, to date, few trials have investigated moderators in obesity prevention trials in young children and even fewer have explored moderators in long-term follow-ups. In the area of early childhood obesity prevention, Yavuz et al. (Reference Yavuz, van Ijzendoorn and Mesman12) in their meta-analysis of fifty studies stated that more obesity prevention studies in early childhood need to perform long-term follow-up in order to better understand the moderators that are associated with sustained effects. Thus, the aim of this study is to assess the effectiveness of INFANT by maternal education and age at 2 and 3·5 years post-intervention.

Methods

Study design and participants

INFANT’s study design, sample selection, intervention and primary outcomes have been described previously(Reference Campbell, Hesketh and Crawford5,Reference Campbell, Lioret and McNaughton6) . Briefly, the control group received usual care from the maternal and child health nurse, whereas the intervention group received usual care plus 6 × 2-h group sessions delivered by a dietician over the 15-month intervention period(Reference Campbell, Hesketh and Crawford5,Reference Campbell, Lioret and McNaughton6) .

A total of 542 children were randomised into the study, with a total of 492 (91 %) completing the trial when the child was 20 months old. All 492 families who completed the programme post-intervention were contacted and asked to partake in the follow-ups when their child was 3·5 and 5 years old(Reference Hesketh, Campbell and Salmon13). At the 2- and 3·5-year follow-ups, a total of 358 and 362 children, respectively, had outcome data for anthropometrics, physical activity, television viewing or diet.

Measures

At both follow-ups, all measures for anthropometrics, physical activity, television viewing and dietary intake were identical.

Anthropometrics

Height and weight were assessed when the children were wearing light clothing using a portable stadiometer (Seca 220/217) and scale (Tanita BWB-800/InnerScan 50), respectively. Height and weight were assessed twice and recorded to the nearest 0·1 cm for height and 10 g for weight. The average of the two measurements was used to calculate BMI. BMI z-scores were then computed using the gender-specific BMI-for-age growth charts from the WHO(14).

Physical activity

The children’s total physical activity (i.e., light, moderate and vigorous intensity) was assessed together using the ActiGraph accelerometer (Model GT1M). The children wore the accelerometers for eight consecutive days above their right hip and were instructed to only take the monitors off for water-based activities (e.g., showering or swimming) and sleeping. The ActiGraphs collected data using 15-s epochs and to be included in the analyses children had to have at least 4 d of valid data where the monitors were worn for a minimum of 7·4 h/d(Reference Hnatiuk, Ridgers and Salmon15). Within this sample, a minimum of 4 d has been found to provide an acceptable reliability estimate (interclass correlation coefficient > 0·70) for total physical activity(Reference Hnatiuk, Ridgers and Salmon15). The cut-points by Janssen et al. (Reference Janssen, Cliff and Reilly16) were used to assess the amount of time spent in physical activity intensities.

Television viewing

The amount of time spent watching television was assessed in the parental questionnaire by the following two questions: ‘on an average weekday (Monday–Friday) how much time does your child spend watching or in front of the television?’ and ‘on an average weekend day (Saturday–Sunday) how much time does your child spend watching or in front of the television?’. These questions have been found to have acceptable reliability (interclass correlation coefficient 0·69, 95 % CI 0·54, 0·80) in a separate sample of Australian parents with a 3- to 5-year-old child(Reference Hinkley, Salmon and Okely17).

Dietary intake

Three multi-pass 24-h dietary recalls(Reference Blanton, Moshfegh and Baer18) over the phone with a parent were used to assess the children’s diet and this procedure has been described previously(Reference Campbell, Lioret and McNaughton6,Reference Cameron, Ball and Hesketh10) . The current analyses utilised the mean daily consumption (in g/d) of fruit (omitting fruit juice), vegetables (omitting potatoes), non-core sweet foods (e.g., bakery products, candy), non-core savoury foods (e.g., potato chips, savoury crackers), non-core drinks (e.g., soda, fruit juice), and water.

Moderator variables

Maternal education

The highest obtained level of education was self-reported in the parental questionnaire when the families were first enrolled in the study when the infant was 4 months old. Maternal education was dichotomised into two groups for these analyses in order to maximise power (university degree v. no university degree).

Maternal age

The mother’s date of birth was self-reported in the parental questionnaire at baseline and maternal age was computed from the information provided. The median age at baseline (32 years) was used to create a dichotomous variable which was used in the analyses.

Statistical analyses

Random effects linear regression models estimated using maximum likelihood, taking into account the cluster-based recruitment of the participants within parent groups, were used to assess the differences in outcomes between the intervention and control group. Due to the fact that some of the outcomes had highly skewed distributions, in all regression models se were bootstrapped (2000 re-samples). Models with BMI z-score as the outcome also included the child’s BMI z-score from baseline as a covariate. However, due to the fact that children were only 4 months of age at baseline, there are no baseline data for any of the dietary variables, television viewing or physical activity. Furthermore, for the models that included physical activity, ActiGraph wear time for that time point was used as a covariate.

In order to evaluate moderation by maternal education and age, models evaluating each outcome comprised terms for maternal education or age and intervention group (i.e., intervention or control group) as well as an interaction term (i.e., maternal education or age × intervention group). The intervention’s effect on the outcome was deemed to be moderated when the interaction term’s coefficient was significantly different to zero(Reference Whisman and McClelland19). As moderator analysis in randomised controlled trial is used for both hypothesis formation and testing(Reference Kraemer, Wilson and Fairburn20) and due to the fact that testing for moderation is affected by the sample distribution and size(Reference Fairchild and MacKinnon21), a P-value of 0·2 was used to indicate significance in the moderation analyses. After the moderation analysis, the direction and strength of the intervention effect were investigated using a linear combinations post-estimation command to obtain the estimated intervention effect for the non-reference level of the moderator. All analyses were conducted with Stata (Release 15, StataCorp LP).

Results

Table 1 presents the baseline characteristics of both the participating mother and child. At the 2- and 3·5-year follow-ups, no significant difference was observed for maternal education between the control group and intervention group (χ 2 = 2·5, P = 0·115 and χ 2 = 1·9, P = 0·173, respectively). No significant differences were observed for the dichotomised variables for maternal age between the intervention and control groups at either the 2-year (χ 2 = 0·1, P = 0·816) or 3·5-year follow-ups (χ 2 =0·1, P = 0·699).

Table 1

Baseline characteristics of the mothers and infants in INFANT

* One mother was missing pre-pregnancy BMI and therefore n 176 and 184 at the 2- and 3·5-year follow-ups, respectively.

Moderating effect of maternal education at the 2- and 3·5-year follow-ups

Table 2 shows the moderation effect of maternal education on the intervention effect at the 2-year follow-up. It was observed that maternal education moderated the intervention effect for the consumption of vegetables (interaction, P = 0·090), water (interaction, P = 0·076) and sweet snacks (interaction, P = 0·079). Vegetable consumption was 30·26 g/d (95 % CI 11·03, 49·50) higher in children in the intervention group with university educated mothers compared to the control group with university educated mothers. The consumption of sweet snacks was –9·30 g/d (95 % CI –14·82, –3·79) lower in children in the intervention group with university educated mothers. Additionally, water intake was 184·63 g/d (95 % CI 47·31, 321·94) higher in children in the intervention group with non-university educated mothers.

Table 2

Assessment of the effect of the intervention in INFANT according to maternal education level at the 2-year follow-up*

MD, mean difference

* Sample sizes: zBMI: total (n 354), low (n 143), high (n 211); dietary variables: total (n 261), low (n 98), high (n 163); television viewing: total (n 343), low (n 135), high (n 208); physical activity: total (n 144), low (n 44), high (n 100).

† High education = university degree; low education = all other.

‡ Mean difference coefficients between the intervention and control group estimated from linear regression analysis.

§ Random effects linear regression models, estimated using maximum likelihood with bootstrapped se, were fitted to compare continuous outcomes between the intervention and control groups. For the models that were measuring the intervention effects on zBMI and physical activity, they were adjusted for baseline zBMI and accelerometer wear time, respectively.

|| Total physical activity (i.e., includes light, moderate and vigorous intensity).

Moderation of the intervention effect by maternal education at the 3·5-year follow-up was observed for zBMI (interaction, P = 0·126) and water intake (interaction, P < 0·001) (Table 3). In stratified analyses, no significant differences were observed between mothers with and without a university education for zBMI or water intake. However, in line with the 2-year follow-up there was a trend for water intake to be greater in children in the intervention group with a non-university educated mothers (MD beta = +114·28 g/d, 95 % CI –38·75, 267·31).

Table 3

Assessment of the effect of the intervention in INFANT according to maternal education level at the 3·5-year follow-up*

MD, mean difference

* Sample sizes: zBMI: total (n 362), low (n 148), high (n 214); dietary variables: total (n 270), low (n 103), high (n 167); television viewing: total (n 352), low (n 142), high (n 210); physical activity: total (n 146), low (n 47), high (n 99).

‡ Mean difference coefficients between the intervention and control group estimated from linear regression analysis.

† High education = university degree; low education = all other.

§ Random effects linear regression models, estimated using maximum likelihood with bootstrapped se, were fitted to compare continuous outcomes between the intervention and control groups. For the models that were measuring the intervention effects on zBMI and physical activity, they were adjusted for baseline zBMI and accelerometer wear time, respectively.

|| Total physical activity (i.e., includes light, moderate and vigorous intensity).

Moderating effect of maternal age at the 2- and 3·5-year follow-ups

Table 4 shows the moderation effect of maternal age on the intervention effect at the 2-year follow-up. It was observed that maternal age moderated the intervention effect for water and sweet snack intake (interaction, P = 0·052 and 0·075, respectively) as well as for physical activity (interaction, P = 0·110). Water consumption was 177·36 g/d (95 % CI 80·54, 274·19) higher in children in the intervention group with mothers aged < 32 years compared to the control group with mothers aged < 32 years. Sweet snack consumption was −9·91 g/d (95 % CI –15·65, –4·17) lower in children in the intervention group with mothers aged ≥ 32 years. No significant intervention effect for physical activity was observed in children of either younger or older mothers; however, there was a trend for children in the intervention group with younger mothers to decrease physical activity (MD beta = –12·95 min/d, 95 % CI –33·92, 8·02) and children in the intervention group with older mothers to increase their physical activity (MD beta = +11·57 min/d, 95 % CI –3·76, 26·91).

Table 4

Assessment of the effect of the intervention in INFANT according to maternal age at the 2-year follow-up*

MD, mean difference

* Sample sizes: zBMI: total (n 354), age < 32 years (n 161), age ≥ 32 years (n 193); dietary variables: total (n 261), < 32 years (n 119), age ≥ 32 years (n 142); television viewing: total (n 343), age < 32 years (n 156), age ≥ 32 years (n 187); physical activity: total (n 144), age < 32 years (n 59), age ≥ 32 years (n 85).

† Mean difference coefficients between the intervention and control group estimated from linear regression analysis.

‡ Random effects linear regression models, estimated using maximum likelihood with bootstrapped se, were fitted to compare continuous outcomes between the intervention and control groups. For the models that were measuring the intervention effects on zBMI and physical activity, they were adjusted for baseline zBMI and accelerometer wear time, respectively.

§ Total physical activity (i.e., includes light, moderate and vigorous intensity).

Moderation of the intervention effect by maternal age at the 3·5-year follow-up was observed for the intake of fruit (interaction, P = 0·046), sweet snacks (interaction, P = 0·136), savoury snacks (interaction, P = 0·003) as well as physical activity (interaction, P = 0·200) (Table 5). Sweet snack intake was –11·54 g/d (95 % CI –20·57, –2·52) lower in children in the intervention group with mothers aged < 32 years compared to the control group with mothers aged < 32 years. With regard to savoury snacks, significant intervention effects in both mothers aged < 32 years and ≥ 32 years were observed in stratified analyses. Children in the intervention group with younger mothers significantly increased their consumption of savoury snacks by 5·56 g/d (95 % CI 0·70, 10·41), whereas children in the intervention group with older mother significantly decreased their consumption of savoury snacks by –5·15 g/d (95 % CI –9·48, –0·81). No significant intervention effect for fruit intake or physical activity was observed in children of either younger or older mothers. However, there was a trend for children in the intervention group with younger mothers to increase their fruit intake (MD beta = +38·40 g/d, 95 % CI –3·29, 80·10) and children in the intervention group with older mothers to decrease their fruit intake (MD beta = −29·37 g/d, 95 % CI –72·19, 13·44). Similar to the 2-year follow-up, there was also a trend for children in the intervention group with younger mothers to decrease their physical activity (MD beta = –9·38 min/d,95 % CI –32·77, 14·00) and children in the intervention group with older mothers to increase their physical activity (MD beta = +10·42 min/d, 95 % CI –4·80, 25·63).

Table 5

Assessment of the effect of the intervention in INFANT according to maternal age at the 3·5-year follow-up*

MD, mean difference

* Sample sizes: zBMI: total (n 362), < 32 years (n 166), ≥ 32 years (n 196); dietary variables: total (n 270), < 32 years (n 123), ≥ 32 years (n 147); television viewing: total (n 352), < 32 years (n 163), ≥ 32 years (n 189); physical activity: total (n 146), < 32 years (n 64), ≥ 32 years (n 82).

† Mean difference coefficients between the intervention and control group estimated from linear regression analysis.

‡ Random effects linear regression models, estimated using maximum likelihood with bootstrapped se, were fitted to compare continuous outcomes between the intervention and control groups. For the models that were measuring the intervention effects on zBMI and physical activity, they were adjusted for baseline zBMI and accelerometer wear time, respectively.

§ Total physical activity (i.e., includes light, moderate and vigorous intensity).

Discussion

The current study examined how maternal education and age moderate the effect of INFANT 2 and 3·5 years post-intervention. At both follow-ups, maternal education and age moderated some of the intervention’s effects; however, clear patterns were not evident at the long-term follow-ups. At the 2-year follow-up, the results regarding maternal education were similar to those obtained post-intervention(Reference Cameron, Ball and Hesketh10). However, these results were not sustained at the 3·5-year follow-up. With regard to maternal age, inconsistent results were obtained across post-intervention(Reference Cameron, Ball and Hesketh10) and both follow-ups. These inconsistent results and the fact that many of the outcomes were not moderated by maternal age or education indicate that INFANT was more-or-less equally effective in children irrespective of their mother’s education level or age.

Maternal education

Similar to what was observed at the end of INFANT(Reference Cameron, Ball and Hesketh10), at the 2-year follow-up children in the intervention group with university educated mothers consumed significantly more vegetables and less sweet snacks compared to their peers whose mother’s had no university education. However, these effects were not sustained at the 3·5-year follow-up. A recent systematic review by Hodder et al. (Reference Hodder, O’Brien and Tzelepis22) found moderate-quality evidence for multi-component interventions to increase vegetable consumption in children under 5 years which included parental involvement in the intervention, a focus on vegetable consumption instead of general nutrition knowledge and nutrition education. The focus on specific messaging and skill development around vegetables and other core and non-core foods that were the focus of INFANT may have supported the vegetable differences observed early in life, but appears to have been more effective for university educated mothers in a child’s early years. The 30 g/d increase in vegetables that was observed in children with university educated mothers at the 2-year follow-up equals approximately half a serving (defined as 75 g), which is about one fifth of the daily serving for a 2-3-year-old child(23). Furthermore, in cross-sectional and longitudinal studies, healthier dietary patterns have also been observed in infants/young children whose mothers have a higher education(Reference Gutierrez-Camacho, Mendez-Sanchez and Klunder-Klunder9,Reference Camara, de Lauzon-Guillain and Heude24) , suggesting this dietary change may be easier to achieve in higher educated groups. Future research should investigate if interventions need to be sustained in order to maintain effects on vegetable consumption.

To the best of our knowledge, no other obesity prevention interventions to date have investigated maternal education as a moderator in long-term follow-ups. However, in addition to INFANT(Reference Cameron, Ball and Hesketh10), two other studies have investigated potential moderation at the end of their interventions. At the intervention conclusion in the Healthy School Start II and ToyBox interventions, no moderation for any of the dietary outcomes or diet quality by parental or maternal education was observed(Reference Nyberg, Norman and Sundblom25,Reference Pinket, De Craemer and Huybrechts26) . The differing results obtained by the current study and these studies could be due to the fact that Healthy School Start II and ToyBox were based in the school and kindergarten setting, whereas INFANT was community based and focused on parents as the agents of change. Furthermore, differences could be attributed to the different age of the children when the interventions were delivered, that is, infants in the current study, 6 years of age in a Healthy School Start II(Reference Nyberg, Norman and Sundblom25) and between 3·5 and 5·5 years in the ToyBox study(Reference Pinket, De Craemer and Huybrechts26). Interventions starting in infancy may be more effective than interventions starting later in life as it has been found that parents are most responsive during infancy(Reference Tickell27). Finally, the differences could also be due to how the dietary data were collected in the studies, with 24-h recalls used in INFANT and FFQ used in the other two studies(Reference Nyberg, Norman and Sundblom25,Reference Pinket, De Craemer and Huybrechts26) . This might have influenced the observed differences between the studies as dietary recalls are a more precise method than FFQ.

In contrast to the other dietary variables, and as observed at the end of INFANT(Reference Cameron, Ball and Hesketh10), a significant increase in water consumption was observed among children with non-university educated mothers. The significant difference was only observed at the 2-year follow-up; however, there was a trend showing that children of non-university educated mothers consumed more water than children with university educated mothers at the 3·5-year follow-up (+114·28 g/d, P = 0·143 v. –29·21 g/d, P = 0·476). In a cross-sectional analysis using the baseline data from ToyBox, children with lower educated mothers consumed significantly less water than children with higher educated mothers in the total sample(Reference Pinket, De Craemer and Maes28). As the results of INFANT indicate that children with non-university educated mothers consumed more water, this could indicate that water intake is a behaviour that can be successfully changed in interventions in lower socio-economic groups. However, this could differ between countries depending on tap water quality and if tap water is perceived as safe. In Australia, tap water is free and accessible and there has been nationwide campaigns from the government (e.g., ‘Choose TAP’(Reference Choose29)) to promote the consumption of tap water. In the European Union, bottled water consumption is prevalent and tap water is overlooked due to perceived notions regarding quality(30). Thus, increasing water consumption in lower socio-economic groups may be more effective in countries such as Australia where tap water is perceived as safe, and its minimal cost becomes a facilitator of change particularly to lower socio-economic groups.

Maternal age

At the 2-year follow-up, children in the intervention group with younger mothers consumed significantly more water than their counterparts with older mothers. This result was also observed post-intervention(Reference Cameron, Ball and Hesketh10); however, it was not sustained at the 3·5-year follow-up. In a recent systematic review by Franse et al. (Reference Franse, Wang and Constant31), maternal age was not found to be a factor associated with water consumption in children. The results observed in the present study could be due to the fact that younger mothers may be more receptive to change as they may not have as strongly developed views regarding parenting-related behaviours as older mothers.

At the 2-year follow-up, sweet snack consumption significantly decreased in children with older mothers. However, at the 3·5-year follow-up the opposite result was found, where sweet snack consumption decreased in children with younger mothers. Furthermore, at the 3·5-year follow-up it was found that the children’s consumption of savoury snacks significantly increased in those with younger mothers and significantly decreased in those with older mothers. While it is unclear why these results were observed, previous studies have found that greater maternal age (≥ 30 years) has been characterised by healthier dietary patterns in infants(Reference Gutierrez-Camacho, Mendez-Sanchez and Klunder-Klunder9). Thus, at the 3·5-year follow-up it could be hypothesised that the younger mothers substituted sweet snacks for savoury snacks and the children’s overall intake of unhealthy snacks did not decrease. However, there was also a trend showing that the intake of fruit in children with younger mothers increased; therefore, sweet snacks could have also been replaced with fruit at this time point.

Strengths and limitations

Strengths of this study include the long-term follow-ups, use of multi-pass 24- h dietary recalls, objective measures of physical activity as well as reliable measure for television viewing. Furthermore, this study reports the moderation analyses of the intervention effects at two long-term follow-ups, which is currently lacking in the contemporary literature regarding obesity prevention interventions in young children. It is important to note that the mother infant dyads excluded from this study at the 2- and 3·5-year follow-ups because of missing data had lower education than the included sample. Another limitation is that change in maternal education level as well as attendance at child care was not accounted for in this study. Additionally, as INFANT was not powered based on assessment of the moderation analyses, the sample size may have been too small to detect some true differences where they actually exist. Larger studies that would permit detection of intervention effects in subgroups, such as maternal education and age, are warranted.

Conclusions

Moderation by maternal education and age were observed for some outcomes; however, clear patterns were not evident at both follow-ups, with little consistency across outcomes. This indicates that INFANT was more-or-less equally effective in children irrespective of their mother’s education level or age, which is important in community-based interventions.

Acknowledgements

Acknowledgements: Not applicable. Financial support: INFANT was funded by National Health and Medical Research Council Project Grants (425801 & 1008879). C.D.N. was supported by a grant from the Swedish Research Council for Health, Working Life and Welfare (2018–01137). A.J.C. was supported by a Heart Foundation Future Leader Fellowship (#36357) from the National Heart Foundation of Australia and is a researcher within a National Health and Medical Research Council (NHMRC) Centre of Research Excellence in Food Retail Environments for Health (RE-FRESH) (APP1152968) (Australia). He has received other funding from the NHMRC, Australian Research Council (ARC), Canadian Institutes of Health Research (CIHR) and the World Health Organization (WHO). K.D.H. was supported by an Australian Research Council Future Fellowship (FT130100637) and Honorary National Heart Foundation of Australia Future Leader Fellowship (100370). K.J.C. and K.D.H. are researchers within a National Health and Medical Research Council (NHMRC) Centre of Research Excellence in the Early Prevention of Obesity in Childhood (1101675). None of the funding sources had involvement in the study design, collection, analysis and interpretation of the data; the writing of the manuscript or the decision to submit the manuscript for publication. Conflict of interest: None. Authorship: K.D.H. and K.J.C. designed the study and are the PI’s of INFANT. C.D.N. conducted the analyses with aid from A.J.C. C.D.N. drafted the manuscript that was reviewed by all co-authors. All co-authors have approved the final version of the manuscript. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Deakin University Human Research Ethics Committee (2007–175). Written informed consent was obtained from all subjects/patients.

References

World Health Organisation (2020) Obesity and overweight. https://www.who.int/en/news-room/fact-sheets/detail/obesity-and-overweight (accessed July 2020).Google Scholar
Huse, O, Hettiarachchi, J, Gearon, E et al. (2018) Obesity in Australia. Obesity Res Clin Pract 12, 29–39.CrossRefGoogle ScholarPubMed
Geserick, M, Vogel, M, Gausche, R et al. (2018) Acceleration of BMI in early childhood and risk of sustained obesity. N Engl J Med 379, 1303–1312.CrossRefGoogle ScholarPubMed
Waters, E, de Silva-Sanigorski, A, Hall, BJ et al. (2011) Interventions for preventing obesity in children. Cochrane Database Syst Rev. Published online: 07 December 2011. doi: 10.1002/14651858.CD001871.pub3.CrossRefGoogle ScholarPubMed
Campbell, K, Hesketh, K, Crawford, D et al. (2008) The Infant feeding activity and nutrition trial (INFANT) an early intervention to prevent childhood obesity: cluster-randomised controlled trial. BMC Public Health 8, 103.CrossRefGoogle ScholarPubMed
Campbell, KJ, Lioret, S, McNaughton, SA et al. (2013) A parent-focused intervention to reduce infant obesity risk behaviors: a randomized trial. Pediatrics 131, 652–660.CrossRefGoogle ScholarPubMed
Hesketh, K, Salmon, J, McNaughton, S et al. (2020) Long-term outcomes (2 and 3·5 years post-intervention) of the INFANT early childhood intervention to improve health behaviors and reduce obesity: cluster randomised controlled trial follow-up. Int J Behav Nutr Phys Act 17, 95.CrossRefGoogle ScholarPubMed
McCrory, C, Leahy, S, Ribeiro, AI et al. (2019) Maternal educational inequalities in measured body mass index trajectories in three European countries. Paediatr Perinat Epidemiol 33, 226–237.CrossRefGoogle ScholarPubMed
Gutierrez-Camacho, C, Mendez-Sanchez, L, Klunder-Klunder, M et al. (2019) Association between sociodemographic factors and dietary patterns in children under 24 months of age: a systematic review. Nutrients 11, 2006.CrossRefGoogle ScholarPubMed
Cameron, AJ, Ball, K, Hesketh, KD et al. (2014) Variation in outcomes of the Melbourne Infant, Feeding, Activity and Nutrition Trial (InFANT) Program according to maternal education and age. Prev Med 58, 58–63.CrossRefGoogle ScholarPubMed
Brown, T, Moore, TH, Hooper, L et al. (2019) Interventions for preventing obesity in children. Cochrane Database Syst Rev 7, Cd001871.Google ScholarPubMed
Yavuz, HM, van Ijzendoorn, MH, Mesman, J et al. (2015) Interventions aimed at reducing obesity in early childhood: a meta-analysis of programs that involve parents. J Child Psychol Psychiatry 56, 677–692.CrossRefGoogle ScholarPubMed
Hesketh, KD, Campbell, K, Salmon, J et al. (2013) The Melbourne Infant Feeding, Activity and Nutrition Trial (InFANT) Program follow-up. Contemp Clin Trials 34, 145–151.CrossRefGoogle ScholarPubMed
WHO Multicentre Growth Reference Study Group (2006) WHO Child Growth Standards based on length/height, weight and age. Acta Paediatr Suppl 450, 76–85.Google Scholar
Hnatiuk, J, Ridgers, ND, Salmon, J et al. (2012) Physical activity levels and patterns of 19-month-old children. Med Sci Sports Exerc 44, 1715–4720.CrossRefGoogle ScholarPubMed
Janssen, X, Cliff, DP, Reilly, JJ et al. (2013) Predictive validity and classification accuracy of ActiGraph energy expenditure equations and cut-points in young children. PLoS One 8, e79124.CrossRefGoogle ScholarPubMed
Hinkley, T, Salmon, J, Okely, AD et al. (2012) The HAPPY study: development and reliability of a parent survey to assess correlates of preschool children’s physical activity. J Sci Med Sport 15, 407–417.CrossRefGoogle ScholarPubMed
Blanton, CA, Moshfegh, AJ, Baer, DJ et al. (2006) The USDA automated multiple-pass method accurately estimates group total energy and nutrient intake. J Nutr 136, 2594–2599.CrossRefGoogle ScholarPubMed
Whisman, MA & McClelland, GH (2005) Designing, testing, and interpreting interactions and moderator effects in family research. J Fam Psychol 19, 111–120.CrossRefGoogle ScholarPubMed
Kraemer, HC, Wilson, GT, Fairburn, CG et al. (2002) Mediators and moderators of treatment effects in randomized clinical trials. Arch Gen Psychiatr 59, 877–883.CrossRefGoogle ScholarPubMed
Fairchild, AJ & MacKinnon, DP (2009) A general model for testing mediation and moderation effects. Prev Sci 10, 87–99.CrossRefGoogle ScholarPubMed
Hodder, RK, O’Brien, KM, Tzelepis, F et al. (2020) Interventions for increasing fruit and vegetable consumption in children aged five years and under. Cochrane Database Syst Rev 5, Cd008552.Google ScholarPubMed
National Health and Medical Research Council Australia (2015) Recommended number of serves for children, adolescents and toddlers. https://www.eatforhealth.gov.au/food-essentials/how-much-do-we-need-each-day/recommended-number-serves-children-adolescents-and (accessed December 2020).Google Scholar
Camara, S, de Lauzon-Guillain, B, Heude, B et al. (2015) Multidimensionality of the relationship between social status and dietary patterns in early childhood: longitudinal results from the French EDEN mother-child cohort. Int J Behav Nutr Phys Act 12, 122.CrossRefGoogle ScholarPubMed
Nyberg, G, Norman, A, Sundblom, E et al. (2016) Effectiveness of a universal parental support programme to promote health behaviours and prevent overweight and obesity in 6-year-old children in disadvantaged areas, the Healthy School Start Study II, a cluster-randomised controlled trial. Int J Behav Nutr Phys Act 13, 4.CrossRefGoogle ScholarPubMed
Pinket, AS, De Craemer, M, Huybrechts, I et al. (2017) Multibehavioural interventions with a focus on specific energy balance-related behaviours can affect diet quality in preschoolers from six European countries: the ToyBox-study. Nutrients 9, 479.CrossRefGoogle ScholarPubMed
Tickell, C (2011) The early years: foundations for life, health and learning. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/180919/DFE-00177-2011.pdf (accessed July 2020).Google Scholar
Pinket, AS, De Craemer, M, Maes, L et al. (2016) Water intake and beverage consumption of pre-schoolers from six European countries and associations with socio-economic status: the ToyBox-study. Public Health Nutr 19, 2315–2325.CrossRefGoogle ScholarPubMed
Choose, TAP (2019) Be a refiller not a landfiller. https://choosetap.com.au/ (accessed July 2020).Google Scholar
Aqua Publica Europea (2019) UNTAPPED Promoting tap water in Europe. https://www.aquapublica.eu/sites/default/files/article/file/Aqua%20Publica%20Europea_Promoting%20Tap%20Water%20in%20Europe_0.pdf (accessed July 2020).Google Scholar
Franse, CB, Wang, L, Constant, F et al. (2019) Factors associated with water consumption among children: a systematic review. Int J Behav Nutr Phys Act 16, 64.CrossRefGoogle ScholarPubMed
Figure 0

Table 1Baseline characteristics of the mothers and infants in INFANT

Figure 1

Table 2Assessment of the effect of the intervention in INFANT according to maternal education level at the 2-year follow-up*

Figure 2

Table 3Assessment of the effect of the intervention in INFANT according to maternal education level at the 3·5-year follow-up*

Figure 3

Table 4Assessment of the effect of the intervention in INFANT according to maternal age at the 2-year follow-up*

Figure 4

Table 5Assessment of the effect of the intervention in INFANT according to maternal age at the 3·5-year follow-up*