The RIFL-P (Brazilian Portuguese Version)

Mothers were given a task to do with their children. The mother sat on a yoga mat and played with her child for five minutes. She was shown pictures of patterns on the shape and color sorter of varying degrees of difficulty and asked to encourage the child to make the patterns with her. The goal was to elicit maternal behaviors and speech related to helping and teaching when the task was slightly too developmentally challenging for the child. Following training on the scale, and after watching the recording once, raters applied codes to each of the 11 items using a five-point Likert scale, ranging from 1 ("Not at all true") to 5 ("Very true"). The task plus coding took around eight minutes.

Parenting Interactions with Children: Checklist of Observations Linked to Outcomes, PICCOLO (Brazilian Portuguese Version)

PICCOLO is a strengths-based checklist of 29 observable behaviors used to assess positive parenting interactions with children aged 10 to 47 months [40]. PICCOLO items are clustered in four domains with seven to eight items per domain: (a) affection (warmth, physical closeness, and positive expressions toward the child); (b) responsiveness (responding sensitively to a child's cues, needs, interests, and behaviors); (c) encouragement (active support of play, exploration, curiosity, skills, and creativity); and (d) teaching (shared conversations and play, cognitive stimulation, explanations, and questions). After watching a 10-min film clip once, raters coded items on a three-point ordinal scale from "Absent", or

not seen, to "Clearly" seen. The task plus coding took around 45 min. Psychometric properties of the original instrument have been found to be strong (IRR ranged between *r* = 0.74 for the responsiveness domain and *r* = 0.80 for the affection domain; Cronbach's α coefficient = 0.91 for the total PICCOLO score (ranging from α's of 0.75 for the responsiveness domain to 0.80 for the teaching domain)) [40], as have those of the PICCOLO Brazilian Portuguese version (IRR ranged between *r* = 0.63 for the encouragement domain and *r* = 0.77 for the affection domain; Cronbach's α coefficient = 0.94 for the total PICCOLO score (ranging from α's of 0.79 for the affection and teaching domains to 0.86 for the responsiveness and encouragement domains)) [41].

#### INTERGROWTH-21st Neurodevelopment Assessment

The INTERGROWTH-21st Neurodevelopment Assessment (INTER-NDA) is a valid and reliable international standardized screening assessment of early child development at 2 years of age [42]. This multidimensional measure provides a comprehensive, rapid assessment of cognition, fine and gross motor skills, language, and positive and negative behavior for children aged 22–30 months. Its 37 items are administered in approximately 15 min using a combination of psychometric techniques, such as direct administration, concurrent observation, and caregiver reports. For all INTER-NDA domains, except for negative behavior, higher scores reflect better outcomes. The INTER-NDA has demonstrated good to acceptable agreement with the Bayley Scales of Infant and Toddler Development, Third Edition [43]. The INTER-NDA is designed to be applied by nonspecialists to high-, middle-, and low-income populations. The INTER-NDA instrument was used to evaluate the development of 3776 children aged 24 months from the 2015 Pelotas Birth Cohort Study in Southern Brazil [35].

### Five Stimulation Markers

This scale was developed by Barros and colleagues [44] to examine the extent to which children experienced a cognitively stimulating environment. Mothers answered no/yes (no = 0, yes = 1) to the following questions about their children's activities in the past week: whether someone read or told a story to the child; whether the child went to a park or playground; whether the child had a story book; whether the child watched TV; whether the child visited anyone's house. The responses for the five items were summed, resulting in a scale from 0 to 5 points. The scale has been found to be related to child development outcomes in Brazil, particularly amongst children whose mothers are low in education [44].

#### 2.4.2. Procedures

Data collection for phase 2 included a home visit to film mother–child interactions (10 min PICCOLO task followed by the 5 min RIFL-P task) when children were around 18 months old. The mothers answered a one-page questionnaire about the five stimulation markers before the home visit was terminated. The filmed interactions (collected at 18 months) were coded and analyzed in relation to demographic data (collected up to two days after delivery, in the perinatal period) of the 2015 Pelotas Birth Cohort Study and developmental data collected when children reached 24 months of age.

#### 2.4.3. Data Coding Procedures

The first (AS) and fourth (TNM) authors, trained by expert coders, scored the RIFL-P videos. Raters were trained until the inter-rater reliability of *r* > 0.80 was reached and then coded the remaining videos independently. Reliability testing was conducted using every fourth film throughout data coding. Following the submission of reliability scores, discrepancies were discussed to minimize rater drift. Scale developers were consulted on three occasions.
