Study design and population
We analyzed data from the 2015 Pelotas (Brazil) Birth Cohort Study. Pelotas is a city in southern Brazil, with around 340,000 inhabitants. All hospital-delivered children live born in Pelotas between 1 January and 31 December 2015, whose mother lived in the urban area of the city, were eligible for the study (8). From the 4,333 eligible live births, 4,275 were assessed at birth (response rate 98.7%). All these children and their mothers were invited to follow-up measures at 3, 12 and 24 months. Further information about the 2015 Pelotas Birth Cohort is available elsewhere (8, 9).
2 years follow-up
In the 2 years follow-up, the interviews were conducted at a research clinic, with a follow-up rate of 95.4%. Mothers answered a questionnaire (updating previous information about the child’s and mother’s health, household characteristics, diseases, child activities, among other topics). All procedures were done by female interviewers trained and retrained during the data collection phase. Written informed consent was obtained for participation in all phases of the study. The 2015 Pelotas birth cohort study was approved by the Research Ethics Committee of the School of Physical Education of Federal University of Pelotas (registration no. 26746414.5.0000.5313).
Infant development
Infant development was assessed when children were between 21–27 months old using the INTER-NDA (INTERGROWTH-21st Neurodevelopment Assessment) (10). It consists of 53 items that are directly administered, concurrently observed and caregiver reported. INTER-NDA was designed to be free from cultural biases and is based upon objective reporting of the child’s performance on cognition, expressive and receptive language, gross and fine motor skills, behavior, attention and social-emotional reactivity. This instrument presents Interclass coefficient ranges between 0.75 and 0.83 (p < 0.001) and a moderate agreement with the Bayley Scales of Infant Development III (k = 0.72, p < 0.001), a well-established child development assessment, measuring cognition, language skills, motor skills and adaptive behavior from 1 to 42 months. The validation tests were perfomed in Oxford, United Kingdom, with eighty-one children aged 23–28 months old (11).
Assessment of children who did not present any serious physical or neural disability precluding them doing the INTER-NDA assessment was conducted by trained interviewers, not experts in neurodevelopment, when children went to the clinic (excluding the phone interviews), or in the family’s home. Age-adjusted scores were calculated for cognition, receptive language, motor domains and global development (mean of the domains combined). The adjustment considers a mean of 0 and a standard deviation of 1.
Childcare attendance
Childcare outside the home was measured using interview data from the 1 and 2-year follow-ups. In the 1 year follow-up, mothers were asked about who took care of their children from birth to 1 year, and children receiving any external childcare in that period were identified. At 2 years, mothers reported whether children were currently attending external childcare. Those who responded “no”, were also asked if the child attended any external childcare since 1 year, although they were not currently in childcare.
For our analyses, childcare attendance reported at ages 1 and 2 years was categorized in three mutually exclusive categories as: a) never went to childcare; b) attended some childcare (either at 1 or at 2 years only); c) always (both 1 and 2 years).
Confounders
Covariates measured in the perinatal interview in this study were: sex (female/male), maternal age (≤ 20, 21–30 and ≥ 31), family income (quintiles), maternal education (0–4, 5–8, 9–11 and 12 + years of schooling), pre-term (< 37 weeks), low birth weight (< 2500 g). Additionally, maternal depression was measured using the Edinburgh Postnatal Depression Scale (EPDS) (12), at the 3-month follow-up, with a cut-off point of ≥ 13 points to indicate the presence of at least moderate depression (13).
In the 2-years follow-up, mothers were asked if they read or told stories to their child (yes/no), if the child visited the house of other people in the past week (yes/no), went to the park or some square in the last week (yes/no), and also if the child participated in the PIM (Primeira Infância Melhor) programme, a government home visitation programme that aims to enhance ECD in vulnerable households with young children.
Statistical analysis
The analyses were conducted using Stata 16.0 in five steps: (a) descriptive analyses of children and their mothers were conducted for the entire cohort and compared with those with valid outcome data; (b) description of the participants, according to childcare attendance; (c) description of the participants according to child development; (d) crude analysis of the relationship between the outcomes and childcare attendance; (e) adjusted analysis of the relationship between the outcomes and childcare attendance.
For the first two stages (a and b), chi-squared tests were used. In steps c, d and e, linear regression models were used. For step e, adjusted analyses were conducted including all covariates at the same time as model I. Model II also included variables related to ECD stimulation. Statistical significance was set at 5%, and 95% confidence intervals are provided.