Time Spent in Different Types of Childcare and Development
Time Spent in Different Types of Childcare and Development
We examined data from the LSAC, a prospective, population-based study of young children's development. Study design and sample information for LSAC are detailed elsewhere. Briefly, LSAC used a two-stage cluster sampling design. The first stage selected Australian postcodes and the second stage sampled children within these postcodes. Postcodes were randomly selected and stratified by state/territory and urban/rural status to ensure a nationally representative sample. The Medicare database, which provides medical and hospital coverage for all Australian permanent residents was then used to randomly select infants born between March 2003 and February 2004 within each stratum. This method identified 8921 infants who were eligible to participate. Of these, 5107 infants were recruited into the study. Our analyses use the first three waves of data when children were 0–1 years, 2–3 years and 4–5 years of age. The study was approved by the Australian Institute of Family Studies Ethics Committee.
Detailed information concerning the time and type of childcare was obtained from face-to-face interviews with the study child's primary caregiver (97% mothers) at 0–1 year and again at 2–3 years. At each time point, the primary caregiver reported whether over the past 1 month the study child had been looked after at regular times during the week by anyone other than the parent living in the home. If the response was 'yes' then the regular type of care, the number of hours each week the child attended, and for how many months the child had been attending this care was ascertained for up to three different types of childcare arrangements.
The types of childcare were: (1) centre-based care (2) family day care (3) nanny or relative and (4) other informal childcare such as care by friends. We excluded a small proportion of children (n=218) enrolled in preschool/kindergarten, as in Australia, they are administratively classified as belonging to the education sector unlike childcare services. We calculated the time spent in each type of childcare as reported by parents at 0–1 year and 2–3 years as: Total Time (TT)=(total hours per week×4.3 weeks in a month)×(total months in childcare). The cumulative time was calculated as the sum of the time spent at the 0–1 year and 2–3 years time period (TT0–1+TT2–3). If a child did not attend childcare, a value of 0 was given. Children who did not attend any type of childcare across the 0–3 years period were classified into the 'primary caregiver only' category and were used as the reference group in all analysis. Our measure of time spent in childcare is similar with research from the US National Institute of Child Health and Human Development study that examined the developmental effects of cumulative time in childcare.
Children's receptive vocabulary at age 4–5 years was directly assessed in the child's home using the well-validated Peabody Picture Vocabulary Test III—LSAC Australian Short Form. Raw scores were scaled according to a Rasch model to enable comparison of scores across waves. Children's externalising and internalising behaviours at age 4–5 years were assessed by the parent and teacher using the Strengths and Difficulties Questionnaire (SDQ). The SDQ contains five subscales, measuring prosocial behaviour, hyperactivity, emotional symptoms, conduct problems and peer problems of five items each. The conduct and hyperactivity subscales of the SDQ were summed to reflect externalising behaviours and the emotional and peer subscales were summed to reflect internalising behaviours. Childhood externalising behaviour problems are expressed in children's outward behaviour reflecting a child's negative reaction to his or her environment. These externalising problems may include disruptive, hyperactive and aggressive behaviours. Internalising behaviour problems primarily affect the child's internal psychological environment rather than the external environment. These internalising problems may include withdrawn, anxious and depressed behaviours. The score for externalising and internalising behaviours ranged in between 0 and 20, respectively, with higher scores indicating higher risk of behavioural problems.
Covariates were identified a priori, on the basis of directed acyclic graphs and from research evidence to be associated with the type and time in childcare and children's cognitive and socioemotional outcomes. Covariates included; the primary caregiver's country of birth, education, employment; annual household income; economic hardship over the last year; geographical remoteness using the Accessibility and Remoteness Index of Australia; family structure; number of siblings; child age, sex and birth weight; parental concern about the child's learning and development; number of children's books in the home; time spent reading to the child; whether the child undertook regular cost activities; the primary caregiver's age, psychological distress using the Kessler 6 scale, and self-reported level of attachment and warmth towards the child.
To address attrition and item non-response in the present study, multiple imputation by chained equations was used to impute missing values. Imputed data sets were generated under the missing at random assumption that uses observed variables in the data set to predict missingness and estimate parameters. The imputation was conducted for the full sample, however, data were analysed only for children who had observed receptive vocabulary scores (n=4066), and parent-reported (n=3646) and teacher-reported (n=3208) externalising and internalising behaviour scores. The imputation model included all outcomes, exposures and covariates. Twenty imputed data sets were generated and the results of the imputed analyses were combined using Rubin's rules. Results using the complete-case data were not substantively different from the imputed analysis and would not change the conclusions of this study; therefore, we report the imputed results.
Subjects and Methods
Study Design and Sample
We examined data from the LSAC, a prospective, population-based study of young children's development. Study design and sample information for LSAC are detailed elsewhere. Briefly, LSAC used a two-stage cluster sampling design. The first stage selected Australian postcodes and the second stage sampled children within these postcodes. Postcodes were randomly selected and stratified by state/territory and urban/rural status to ensure a nationally representative sample. The Medicare database, which provides medical and hospital coverage for all Australian permanent residents was then used to randomly select infants born between March 2003 and February 2004 within each stratum. This method identified 8921 infants who were eligible to participate. Of these, 5107 infants were recruited into the study. Our analyses use the first three waves of data when children were 0–1 years, 2–3 years and 4–5 years of age. The study was approved by the Australian Institute of Family Studies Ethics Committee.
Childcare Measures
Detailed information concerning the time and type of childcare was obtained from face-to-face interviews with the study child's primary caregiver (97% mothers) at 0–1 year and again at 2–3 years. At each time point, the primary caregiver reported whether over the past 1 month the study child had been looked after at regular times during the week by anyone other than the parent living in the home. If the response was 'yes' then the regular type of care, the number of hours each week the child attended, and for how many months the child had been attending this care was ascertained for up to three different types of childcare arrangements.
The types of childcare were: (1) centre-based care (2) family day care (3) nanny or relative and (4) other informal childcare such as care by friends. We excluded a small proportion of children (n=218) enrolled in preschool/kindergarten, as in Australia, they are administratively classified as belonging to the education sector unlike childcare services. We calculated the time spent in each type of childcare as reported by parents at 0–1 year and 2–3 years as: Total Time (TT)=(total hours per week×4.3 weeks in a month)×(total months in childcare). The cumulative time was calculated as the sum of the time spent at the 0–1 year and 2–3 years time period (TT0–1+TT2–3). If a child did not attend childcare, a value of 0 was given. Children who did not attend any type of childcare across the 0–3 years period were classified into the 'primary caregiver only' category and were used as the reference group in all analysis. Our measure of time spent in childcare is similar with research from the US National Institute of Child Health and Human Development study that examined the developmental effects of cumulative time in childcare.
Child Outcomes
Children's receptive vocabulary at age 4–5 years was directly assessed in the child's home using the well-validated Peabody Picture Vocabulary Test III—LSAC Australian Short Form. Raw scores were scaled according to a Rasch model to enable comparison of scores across waves. Children's externalising and internalising behaviours at age 4–5 years were assessed by the parent and teacher using the Strengths and Difficulties Questionnaire (SDQ). The SDQ contains five subscales, measuring prosocial behaviour, hyperactivity, emotional symptoms, conduct problems and peer problems of five items each. The conduct and hyperactivity subscales of the SDQ were summed to reflect externalising behaviours and the emotional and peer subscales were summed to reflect internalising behaviours. Childhood externalising behaviour problems are expressed in children's outward behaviour reflecting a child's negative reaction to his or her environment. These externalising problems may include disruptive, hyperactive and aggressive behaviours. Internalising behaviour problems primarily affect the child's internal psychological environment rather than the external environment. These internalising problems may include withdrawn, anxious and depressed behaviours. The score for externalising and internalising behaviours ranged in between 0 and 20, respectively, with higher scores indicating higher risk of behavioural problems.
Covariates
Covariates were identified a priori, on the basis of directed acyclic graphs and from research evidence to be associated with the type and time in childcare and children's cognitive and socioemotional outcomes. Covariates included; the primary caregiver's country of birth, education, employment; annual household income; economic hardship over the last year; geographical remoteness using the Accessibility and Remoteness Index of Australia; family structure; number of siblings; child age, sex and birth weight; parental concern about the child's learning and development; number of children's books in the home; time spent reading to the child; whether the child undertook regular cost activities; the primary caregiver's age, psychological distress using the Kessler 6 scale, and self-reported level of attachment and warmth towards the child.
Multiple Imputation
To address attrition and item non-response in the present study, multiple imputation by chained equations was used to impute missing values. Imputed data sets were generated under the missing at random assumption that uses observed variables in the data set to predict missingness and estimate parameters. The imputation was conducted for the full sample, however, data were analysed only for children who had observed receptive vocabulary scores (n=4066), and parent-reported (n=3646) and teacher-reported (n=3208) externalising and internalising behaviour scores. The imputation model included all outcomes, exposures and covariates. Twenty imputed data sets were generated and the results of the imputed analyses were combined using Rubin's rules. Results using the complete-case data were not substantively different from the imputed analysis and would not change the conclusions of this study; therefore, we report the imputed results.
Source...