This parallel randomised controlled trial was conducted in south west Sydney, Australia, from June 2007 to December 2010. A detailed research protocol has been published elsewhere.25
Participants and recruitment
Research assistants gave pregnant women attending antenatal clinics a letter of invitation and information about the study. Women were eligible for the trial if they were aged 16 and over, expecting their first child, between weeks 24 and 34 of pregnancy, able to communicate in English, and lived in the local area. The recruitment took almost 12 months to complete. Of 2700 mothers who were approached by research assistants, 780 were eligible. We could not establish the eligibility of the others as they declined to participate when approached and we were not able to obtain further information.
Once eligibility was established and consent obtained, women were asked to complete a registration form to allow the nurses to make arrangements for baseline data collection. One of four research nurses conducted the baseline assessments at the woman’s home before randomisation. Because of resource (research staff) constraints we were not able to complete the baseline assessment and randomisation for all participating mothers as planned before they gave birth. Four hundred and nine women were interviewed before birth and 258 after birth.
The sample size calculation was based on the primary outcome, BMI, which was assumed to have a SD of 1.5. To have 80% power to detect a difference in mean BMI of 0.25 units between the groups at age 2 at the two sided 5% significance level, we needed a sample size of 252 per group. To allow for an estimated 20% drop out we aimed to recruit 630 first time mothers.
Random allocation was concealed by sequentially numbered, sealed opaque envelopes containing the group allocation, which was determined by a computer generated random number with a block size of 50 with a 1:1 allocation ratio. Immediately after baseline data collection, the nurse opened the sealed envelope and informed the mother of her group.
Two research assistants not involved in the implementation of the intervention collected outcome data in the woman’s home. The data collectors and data entry staff were blinded to treatment allocation, but the participating mothers were not blinded.
This staged and home based intervention was based on home visiting programmes that have been established as effective interventions for improving the health and wellbeing of parents and children from vulnerable and disadvantages families.27
It was developed through a pilot study30
and guided by health promotion principles. A description of the programme and all intervention resources developed for this study is available online (www.healthybeginnings.net.au/
). The intervention resources promoting breast feeding, appropriate timing of introduction of solids, “tummy time,” and active play, as well as family nutrition and physical activity, were based on various Australian National Guidelines.31
The key intervention messages included (also see appendix):
- Breast is best
- No solids for me until 6 months
- I eat a variety of fruit and vegetables every day
- Only water in my cup
- I am part of an active family.
Four community nurses were recruited and trained to ensure consistency of delivering the intervention. The nurse visited participating families in the intervention group eight times at home, once at 30-36 weeks’ gestation and seven times after the birth (at 1, 3, 5, 9, 12, 18 and 24 months). The timing of the visits corresponds to milestones in early childhood development. At each visit, the nurse spent about one to two hours with the mother and infant. The nurse not only taught the mother specific skills and knowledge in relation to healthy infant feeding practices and active play (the key messages of the programme) but also discussed any issues and concerns raised by the mother. A visit checklist with standard minimum information plus additional discussion points for each key area plus appropriate resources to support each key message was developed. Four key areas included infant feeding practices, child nutrition and active play, family physical activity and nutrition, and social support. After each visit, the nurses documented all aspects of their visits with the participating families and provided regular reports to the investigators on questions and issues arising.
Families in both the control and intervention group received the usual childhood nursing service from community health service nurses. All new mothers in the state of New South Wales receive at least one nurse visit for general support at home. Some vulnerable families are offered multiple home visits. To maximise the retention rate in this study, we posted home safety promotion materials to women in the control group at six and 12 months.
Primary and secondary outcomes
The primary outcome at 24 months was the child’s anthropometric measures for BMI. Secondary outcomes were eating habits (intake of fruit and vegetables, consumption of chips and snacks, and having a meal in front of the TV), time spent watching TV, and active play time, as well as the mothers’ dietary behaviours, time spent watching TV, and physical activity. Measurements were undertaken during a home interview with the mother. The primary outcomes at six months, including exclusive breast feeding and timing of introduction of solids and “tummy time,” were collected by a telephone interview, and those at 12 months, including breastfeeding, cup use, bottle at bedtime, and having food as a reward, were collected by research assistants in the home. These have been reported elsewhere.26
—We calculated children’s BMI (weight (kg)/length (m)2
) at age 2. BMI is one of the best measures of change in adiposity in growing children.33
We also categorised BMI as overweight, obese, or not overweight or obese based on internationally accepted criteria.34
The healthy BMI ranges for children aged 2 are 14.12-18.41 for boys and 13.90-18.02 for girls.33
Length—A research assistant took two measurements of length with the child in a supine position on a level floor (with a SECA 210 Infant Measuring Mat, Hamburg, Germany) and recorded it to the nearest 0.1 cm; a third measure was taken if the first two measurements differed by 0.5 cm or more, and the mean of these two or three values was calculated.
Weight—The research assistant used digital scales (Tanita model 1583 Baby Scale, Tokyo, Japan) to weigh children in light clothes and no shoes. The measures were recorded to the nearest 0.1 kg.
—Mothers reported their child’s eating habits using a short food frequency questionnaire that was specifically designed to assess children’s eating habits, the validity and reliability of which were tested before this study.35
The questionnaire asked about servings of fruit and vegetables; frequency of eating snack foods (biscuits, cakes, donuts, muesli bars), potato crisps and drinking cups of soft drinks/cordials, juice, and water; and frequency of eating in front of the TV and having food as reward.
TV viewing time and outdoor play time
—Mothers reported the total time their child spent watching TV or outdoor play time each day in a usual week using a set of validated questions.36
Mothers’ nutrition and physical activity
—Mothers’ own dietary behaviours and physical activity were assessed with questions sourced from the New South Wales Health Survey Program37
in New South Wales, Australia. These questions have been validated in an adult population and are widely used in population health surveys in New South Wales. We have reported mothers’ dietary behaviours during pregnancy in details elsewhere.38
Other outcomes, including consumption of “junk food” by mothers and children, were assessed with questions from the validated questionnaire and the New South Wales Adult Health Surveys.35
At baseline we collected sociodemographic data including age, employment status, education level, marital status, language spoken at home, and country of birth of mothers, using the standard New South Wales Health Survey questions.31
For most analyses, BMI was used as a continuous variable. We also categorised children at age 2 as overweight/obese or not, based on the age standardised cut points for BMI recommended by the International Obesity Taskforce.33
We examined the outcome variables including eating habits/dietary behaviours, physical activity/outdoor play, and TV viewing time for their distribution, then recategorised them dichotomously according to the median intake of vegetables or fruit or the national guidelines for physical activity, as appropriate. For example, the National Physical Activity Guidelines recommend that screen time for children aged 2-5 is 60 minutes a day maximum and that adults spend at least 150 minutes in moderate intensity physical activity each week.32
Median intake a day was two servings of vegetables or fruit, as reported by participating mothers at the baseline.38
We performed a complete case analysis and intention to treat analyses for each outcome. For the complete case analysis, we compared outcomes at 24 months between intervention and control groups using the two sample t test for the continuous outcome (BMI) and Pearson’s χ2 test for categorical data. We also calculated risk differences with 95% confidence intervals. All P values are two sided and significance was set at 5%.
For intention to treat analyses, we used multiple imputation by chained equations to impute missing values. We imputed the BMI values that were missing for 14 infants who remained in the study at 24 months. We also imputed all missing values of BMI and the other outcomes at 24 months for a full intention to treat analysis of all 667 randomised participants. In both cases the imputation model predicting BMI was based on all plausible observed values of BMI and covariates at baseline and at 6, 12 and 24 months’ follow-up. The imputation models for the binary outcomes were logistic regression models containing exclusive breast feeding, introduction of solid food regularly, and daily practice of “tummy time” at 6 months, and being given food for reward and drinking from a cup at 12 and 24 months. We used 20 imputations each time, which gave a relative efficiency of 99%. We then calculated pooled estimates of the difference in mean BMI and of the odds ratio of having each of the binary outcomes for those in the intervention group compared with the control group. All analyses were performed with Stata version 10 (StataCorp, College Station, TX).